School of Computing and Engineering

Pearson Higher National in Computing

Assignment Brief - BTEC (RQF)
Academic Year 22-23

Student Name: Azwa Danish Class: HND Y1 & Y2

Assessor Name: Unaiza Tallal Issue Date: 05-May-23
Unit No. and Title: Unit 26: Machine Learning
Assignment No. and Title: Assignment 02: Applications and Challenges in Business
Formative Submission 29-May-23 Submitted on:
Date:
Summative Submission 19-Jun-23 Submitted on:
Date:
Resubmission Date: 03-Jul-23 Submitted on:
IV of Brief Name: Dr Arfan Nagra Date of IV: 21-May-23

Submission Format
This assignment must be typed in full, but all work must be shown to demonstrate your
understanding of the tasks.
This assignment is to be submitted directly to Moodle on the date shown above.
Unit Learning Outcomes
Develop a machine learning application using an appropriate programming
LO3
language or machine learning tool for solving a real-world problem.
Evaluate the outcome or the result of the application to determine the
LO4
effectiveness of the learning algorithm used in the application.

Student Declaration

I certify that the assignment submission is entirely my work and I fully understand the
consequences of plagiarism. I understand that making a false declaration is a form of malpractice.

Student Signature: Date:

Pearson BTEC HN RQF Assignment Brief and Student Declaration Form Template Academic Year 2022/23
Attach this section to the front of your completed work

Pearson BTEC HN RQF Assignment Brief and Student Declaration Form Template Academic Year 2022/23
Assignment Brief
Vocational Scenario
You are a data analyst for a large retail company that operates both online and offline stores. The
company is interested in exploring the potential of machine learning to improve its operations and
increase profitability. Your task is to produce a report that outlines the applications of machine
learning in retail, the challenges associated with implementing machine learning, and the potential
benefits of using machine learning for the company's operations.
Assignment Activity and Guidance

Your report should cover the following topics:

1. An introduction to machine learning and its applications in business.

2. The different types of machine learning algorithms, including supervised, unsupervised, and
reinforcement learning, and their use cases in retail.
3. The challenges associated with implementing machine learning in a retail environment,
include data quality, privacy, and ethical concerns.
4. Case studies of companies that have successfully implemented machine learning in retail,
including their approach, benefits, and challenges.
5. Recommendations for the company on how to implement machine learning, including
identifying relevant use cases, selecting appropriate algorithms, and addressing data quality
and privacy concerns.

Recommended Resources

*Please access HN Global for additional resources support and reading for this unit. For further
guidance and support on report writing please refer to the Study Skills Unit on HN Global. Link
to www.highernationals.com

Pearson BTEC HN RQF Assignment Brief and Student Declaration Form Template Academic Year 2022/23
Learning Outcomes and Assessment Criterion
Pass Merit Distinction
LO3 Develop a machine learning application using an LO3 & LO4
appropriate programming language or machine learning
tool for solving a real-world problem D2. Critically evaluate the
P5. Chose an appropriate M3. Test the machine implemented learning
learning problem and learning application using solution and it's
prepare the training and a range of test data and effectiveness in meeting
test data sets in order to explain each stages of this end user requirements.
implement a machine activity.
learning solution.
P6. Implement a machine
learning solution with a
suitable machine learning
algorithm and
demonstrate the outcome.

Pearson BTEC HN RQF Assignment Brief and Student Declaration Form Template Academic Year 2022/23
LO4 Evaluate the outcome or the result of the application to
determine the effectiveness of the learning algorithm used in the
application
P7. Discuss whether the
result is balanced, underfitting
M4. Evaluate the effectiveness
or over-fitting.
of the learning algorithm used
P8. Analyse the result of the
in
application to determine
the application.
the effectiveness of the
algorithm

Pearson BTEC HN RQF Assignment Brief and Student Declaration Form Template Academic Year 2022/23
Formative Feedback

Student Name:

Assessor Signature: Date:

Student Signature: Date:

Summative Feedback

Student Name:

Assignment Grade: Choose an item. Assessors Signature:

Resubmission Feedback

Final Assignment Assessors Signature:

Choose an item.
Grade: Student Signature:
Please note that grade decisions are provisional. They are only confirmed once internal and external moderation
agreed upon at the Examination Board.
IVs Signature and Date:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Table of Contents
Assignment Brief....................................................................................................................................................3
Formative Feedback.................................................................................................................................................6
Summative Feedback..............................................................................................................................................6
Introduction To Machine Learning.........................................................................................................................7
Machine Learning Transformations in Retail Giants: Unveiling Success Stories...................................................13
P5. Choose an appropriate learning problem and prepare the training and test data sets in order to implement
a machine learning solution.................................................................................................................................17
P6. Implement a machine learning solution with a suitable machine learning algorithm and demonstrate the
outcome...............................................................................................................................................................23
P7. Discuss whether the result is balanced, underfitting or overfitting................................................................45
&........................................................................................................................................................................... 45
P8. Analyse the result of the application to determine the effectiveness of the algorithm.................................45
M3. Test the machine learning application using a range of test data and explain each stages of this activity.. .51
M4. Evaluate the effectiveness of the learning algorithm used in the application...............................................56
D2. Critically evaluate the implemented learning solution and its effectiveness in meeting end-user
requirements........................................................................................................................................................57

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Introduction To Machine Learning
Machine Learning (ML) stands at the forefront of a technological revolution, reshaping industries and
profoundly influencing our daily lives. In our increasingly digital world, the generation of vast datasets
poses both challenges and opportunities. ML, a subset of Artificial Intelligence (AI), equips us with
potent tools to distill valuable insights from these immense data volumes, empowering businesses,
researchers, and individuals to make well-founded decisions. This essay delves into the foundational
principles of Machine Learning, offering insights into its core concepts and practical applications
through a real-time case study.

Essentially, Machine Learning revolves around crafting algorithms and models that enable computers
to autonomously analyze data without explicit programming. Utilizing advanced statistical techniques,
ML algorithms automatically discern patterns, make predictions, and inform data-driven decisions.
The heart of Machine Learning lies in training these models with historical data, enabling them to
generalize and accurately forecast outcomes on entirely new, unseen data sets.

 Real-Time Case Study: Fraud Detection in Financial Transactions:

We examine a current case study of financial transaction fraud detection to illustrate how machine
learning is used in practise. Fraudulent conduct in banking and internet transactions provide serious
threats to both people and businesses. Given the enormous volume and complexity of transactional
data, detecting and preventing such acts is a difficult undertaking. But machine learning algorithms
have shown to be very good at spotting trends and abnormalities that point to fraudulent behaviour.

Case Study:

In this case study, a financial institution used a real-time transactional data analysis system that was
based on machine learning to detect fraud. The algorithm was trained using historical data that
included details on both known fraudulent behaviours and information on lawful transactions. The ML
model developed the ability to recognise suspicious behaviours with a high degree of accuracy by
studying a wide range of parameters such transaction amount, location, and user behaviour.

The implementation of the Machine Learning system yielded several benefits.

 It greatly decreased the false positive rate, preventing legal transactions from being mistakenly
labelled as fraudulent. This improved customer service and reduced unneeded annoyance.
 The system quickly alerted the institution to questionable transactions, allowing it to take timely
action and minimising financial losses brought on by fraudulent activity.
Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
 The system grew and responded to new fraud strategies as it continuously learnt from new data,
making it a strong and reliable solution.

The ability of merchants to analyse massive volumes of data and derive insightful conclusions thanks
to machine learning algorithms has revolutionised the retail sector. These algorithms may analyse and
process data to produce precise forecasts, boost decision-making, improve consumer experiences,
and optimise various retail operations.We will investigate a few well-known machine learning
techniques and the retail industry in which they are used.

1. Recommendation Systems:

To make suggestions for goods or services to customers based on their preferences and actions,
recommendation systems are frequently employed in retail and e-commerce. These programmes
make use of collaborative filtering, content-based filtering, and hybrid strategies, among other
machine learning techniques. Personalised recommendations can be provided by recommendation
systems by analysing consumer information, purchase history, browsing habits, and social
interactions. This boosts customer happiness and boosts sales.

2. Demand Forecasting:

For inventory management, pricing strategies, and supply chain optimisation, accurate demand
forecasting is essential. Time series analysis, regression models, and deep learning are examples of
machine learning algorithms that can forecast future demand based on past data, seasonality, trends,
promotions, and outside influences. Retailers may optimize inventory levels, prevent stockouts, lessen
overstocking, and make wise judgements about pricing and promotions by estimating demand.

3. Customer Segmentation:

Client segmentation is the process of breaking down a client base into different groups according to a
variety of factors, including demographics, spending patterns, preferences, and loyalty. Customer
segments can be automatically identified using machine learning algorithms that use clustering
methods like k-means, hierarchical clustering, and Gaussian mixture models. Retailers can customise
marketing campaigns, customise product assortments, and deliver tailored offers to maximise
consumer engagement and retention by understanding distinct client segments.

4. Fraud Detection:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Fraudulent activities, such as identity theft, payment fraud, and return fraud, can significantly impact
retail businesses. Machine learning algorithms, including anomaly detection, neural networks, and
decision trees, can analyze vast amounts of data and detect unusual patterns or fraudulent behaviors.
By implementing fraud detection algorithms, retailers can identify and mitigate potential risks, protect
customer data, and minimize financial losses.

5. Price Optimization:

A key component of retail operations is figuring out the best pricing strategy. In order to dynamically
optimise prices, machine learning algorithms including dynamic pricing models, reinforcement
learning, and regression analysis can examine market trends, rival pricing, demand elasticity, and
customer behaviour. Retailers are able to set pricing that are competitive, increase revenue, and react
rapidly to market changes by utilising these algorithms.

6. Supply Chain Management:

For merchants to ensure timely delivery, decrease costs, and minimise inventory holding, effective
supply chain management is critical. Supply chain processes can be optimised using machine learning
techniques such as demand forecasting, inventory optimisation, and route optimisation. To streamline
the supply chain process and enhance overall efficiency, these algorithms take into account a variety
of elements such as demand patterns, lead times, transportation costs, and warehousing capacities.

The Transformative Power of Machine Learning: Applications Across Diverse Industries

Machine learning, a subset of artificial intelligence, has transformed how businesses function and
make choices. Machine learning enables organisations to extract important insights, automate
processes, and optimise outcomes by harnessing massive volumes of data and complex algorithms.
This paper investigates the different uses of machine learning in numerous industries, highlighting
unique cases that demonstrate its transformational potential.

 Healthcare Sector:
Machine learning has emerged as a game-changer in healthcare, offering improved diagnostics,
personalized treatments, and efficient resource allocation.

1. Disease Diagnosis:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Machine learning algorithms have the remarkable ability to analyze medical images such as X-rays or
MRIs, surpassing the accuracy of human experts in detecting diseases like cancer or pinpointing
abnormalities. An algorithm created by Google's DeepMind can diagnose diabetic retinopathy by
examining retinal images, facilitating early detection and prompt treatment.

2. Predictive Analytics:

Machine learning algorithms can anticipate patient outcomes, allowing healthcare providers to
intervene earlier and more effectively. Algorithms can detect patients at risk of acquiring illnesses
such as sepsis or readmission by analysing patient data such as vital signs and medical history. This
enables healthcare practitioners to better allocate resources and prevent unpleasant effects.

 Business and Marketing:

Machine learning has changed the business environment by optimising processes, improving
consumer experiences, and influencing strategic decision-making.

1. Customer Relationship Management (CRM):

Machine learning algorithms can forecast buying trends and preferences by analysing client data such
as previous purchases, browsing behaviour, and demographic information. This allows businesses to
deliver personalised marketing campaigns and recommendations, increasing client happiness and
sales.

Amazon's recommendation system is a prime example of machine learning in action.

2. Fraud Detection:

Machine learning is used by financial organisations to detect fraudulent activity by analysing massive
amounts of transactional data. Algorithms can detect aberrant patterns and warn suspicious
transactions, saving financial losses.

PayPal utilizes machine learning to safeguard customer transactions, reducing fraudulent activities
significantly.

 Transportation and Logistics:

Machine learning has revolutionized the transportation and logistics industry, optimizing routes,
enhancing safety, and improving operational efficiency.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
1. Route Optimization:

To optimise delivery routes, machine learning algorithms can analyse real-time traffic data, weather
conditions, and historical patterns. Travel time, fuel usage, and prices are all reduced as a result.

Machine learning is used by companies such as UPS to design efficient delivery routes, which
improves overall service quality.

2. Predictive Maintenance:

Machine learning algorithms can forecast equipment breakdowns by analysing sensor data from
vehicles and machinery. This allows for preventive maintenance, reducing downtime and increasing
operational efficiency.

For example, airlines employ machine learning algorithms to detect potential failures in aircraft
engines, ensuring passenger safety and reducing maintenance costs.

 Education and E-Learning:

Machine learning has the potential to transform education, personalizing learning experiences, and
providing valuable insights to educators.

1. Adaptive Learning:

To build personalised learning paths, machine learning algorithms can analyse students' learning
patterns, strengths, and shortcomings. Machine learning improves student engagement and academic

performance by adapting curriculum and assessments to individual requirements.

Duolingo, an AI-powered language-learning platform, employs machine learning to adapt lessons to

learners' proficiency levels.

2. Intelligent Tutoring Systems:

Carnegie Learning's AI-based math tutoring system is a notable example.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Machine learning can provide intelligent teaching systems that deliver individualised feedback and
suggestions to pupils. By assessing student responses, spotting misconceptions, and giving tailored
explanations, these technologies mimic the function of a human instructor.

Machine learning has created unprecedented prospects for growth and innovation across industries.
From healthcare to business, transportation to education, its applications have altered the way
organisations run and service their constituents. With continual breakthroughs in algorithms and data
availability, the significance of machine learning is sure to rise, establishing a future in which
intelligent decision-making becomes the norm.

Machine Learning Transformations in Retail Giants: Unveiling Success Stories

Machine learning has become a strong tool for driving growth, optimising operations, and improving
consumer experiences in the ever-changing retail business. This essay digs into successful machine
learning applications in retail behemoths such as Zara and Alibaba, providing original instances that
demonstrate the transformative influence of machine learning.
 Demand Forecasting and Inventory Management:
Accurate demand forecasting and efficient inventory management are crucial for retailers to optimize
stock levels, minimize costs, and meet customer expectations.

1. Zara's Fast Fashion Revolution:

Zara, known for its quick fashion approach, uses machine intelligence to precisely estimate client
demand. Zara's algorithms enable dynamic inventory replenishment and production planning by
analysing past sales data, current fashion trends, and external factors such as weather and social
media attitude. This guarantees that trendy products are available and avoids the danger of surplus
inventory.

2. Alibaba's Intelligent Supply Chain:

Alibaba, the e-commerce behemoth, optimises its supply chain processes using machine learning
algorithms. Alibaba's technologies forecast product demand across many locations and dynamically
alter inventory levels by taking into account factors such as order history, consumer behaviour, and
transportation data. This method simplifies logistics, shortens delivery times, and increases customer
satisfaction.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Personalized Recommendations and Customer Engagement:
Machine learning has revolutionized personalized marketing, enabling retailers to deliver tailored
recommendations and targeted promotions to customers.

1. Amazon's Recommendation Engine:

Amazon's machine learning-powered recommendation engine has become synonymous with retail.
The algorithm proposes comparable things to customers based on their browsing behaviour, purchase
history, and product similarities, boosting cross-selling and up-selling opportunities. This customised
strategy increases consumer engagement and sales.
2. Alibaba's Smart Stores:

Alibaba's "New Retail" idea combines online and offline purchasing experiences. Alibaba's smart
stores use machine learning to collect and analyse customer data in real time, including purchase
history and preferences. This enables personalised recommendations, interactive displays, and
seamless checkout experiences, blurring the barriers between physical and digital shopping.

1. Walmart's Fraud Detection:

To detect fraudulent transactions and activities, Walmart employs machine learning algorithms. The
system identifies trends that signal probable fraud by analysing massive volumes of transactional
data, assisting Walmart in mitigating financial losses and protecting the sensitive information of its
consumers.

2. Alibaba's Facial Recognition Payments:

"Smile to Pay," Alibaba's unique payment system, makes use of machine learning-based facial
recognition technology. Customers can finish purchases just by smiling into a camera. The system
compares the customer's facial features to data recorded in the system, enabling secure and
convenient payments while lowering the risk of identity theft and fraud.

Impact of Machine Learning in Social Media Analysis for Retail Marketing

Social media has become a massive store of customer opinions, interests, and behaviours in the
digital age. Retailers can extract important insights from social media data by leveraging the power of

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
machine learning, allowing them to better understand their target audience, optimise marketing
strategies, and drive business growth. This section looks at how machine learning is redefining social
media analysis in the retail business and changing the way marketers use customer insights.

1. Sentiment Analysis: Understanding Customer Perceptions

Machine learning algorithms can assess customer sentiment towards products, companies, or specific
campaigns by analysing social media postings, comments, and reviews. Retailers can swiftly discover
consumer opinions and alter their marketing tactics by using natural language processing technology.
To boost brand reputation and customer happiness, merchants can use sentiment analysis to discover
areas for development, address customer issues, and capitalise on good feedback.

2. Social Listening: Identifying Emerging Trends and Influencers

Machine learning algorithms can filter through massive volumes of social media data to find upcoming
retail trends, topics, and influencers. Retailers can remain ahead of the curve and change their
product offers to match changing consumer expectations by monitoring conversations and tracking
keywords. Social listening also aids in the identification of prominent voices that can be used to
facilitate brand partnerships, collaborations, or focused marketing initiatives.

3. Customer Segmentation: Tailoring Marketing Efforts

Social media users can be segmented using machine learning algorithms based on their interests,
demographics, and behaviour patterns. Retailers can adjust their marketing efforts to specific target
demographics by building different consumer segments. With this level of personalisation, businesses
can present more relevant information, offers, and recommendations, resulting in increased
consumer engagement and conversion rates.

4. Social Media Advertising Optimization: Maximizing ROI

To optimise advertising campaigns, machine learning algorithms can analyse past social media
advertising data, such as click-through rates, conversions, and engagement metrics. Retailers can use
this data to find the most effective ad locations, target the appropriate audience segments, and
allocate advertising expenditures more efficiently. This data-driven approach maximises ROI and
ensures that marketing efforts are directed towards the platforms and techniques that produce the
best outcomes.

5. Influencer Marketing: Identifying the Right Influencers

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Social media data can be analysed by machine learning algorithms to find influencers who are
compatible with a retailer's brand values and target market. Machine learning assists shops in
selecting the most appropriate influencers for collaborations or ambassador programmes by looking
at variables like follower demographics, engagement rates, and content relevancy. This data-driven
strategy guarantees that businesses can form genuine and successful alliances with influencers to
connect with their target audience.

P5. Choose an appropriate learning problem and prepare the training and test data sets in order to
implement a machine learning solution.

Leveraging Machine Learning for Enhanced Retail Operations: Applications, Challenges, and
Benefits

Problem Statement:

The company, which employs data analysts, works for a sizable retailer with both online and physical
stores, and it is aware of the potential of machine learning to boost productivity and profitability. The
current job is to provide a thorough report explaining the uses of machine learning in the retail sector,
addressing the implementation obstacles, and showcasing the possible advantages it might offer to
the business's operations.

Scenario Explanation:

In the competitive landscape of the retail industry, companies are continually striving to enhance their
operations for a competitive edge and to meet customer needs effectively. Recognizing the pivotal
role of machine learning in leveraging data-driven insights, the retail company acknowledges its
significance in informed decision-making.

This report delves into diverse machine learning applications within retail, spanning personalized
marketing, demand forecasting, inventory optimization, fraud detection, and customer segmentation.
These applications empower the company to customize marketing strategies, predict customer
demands, manage stock levels efficiently, identify fraudulent activities, and target specific customer
segments precisely.

Nevertheless, integrating machine learning into retail operations presents challenges. The report
addresses these hurdles, including issues related to data quality, integration with existing systems, the
necessity for skilled personnel, and ethical considerations regarding customer data privacy and
security. Understanding these challenges is imperative for the company to devise a successful

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
machine learning implementation strategy. The report underscores the advantages of embracing
machine learning, such as heightened operational efficiency, increased sales, improved customer
satisfaction, cost reduction through optimized inventory management, and the ability to make real-
time data-driven decisions.

Through a comprehensive analysis of machine learning applications, challenges, and benefits tailored
to the retail company's context, this report furnishes invaluable insights to stakeholders. It serves as a
guiding resource, enabling the company to harness machine learning effectively, fostering operational
excellence, and driving profitability in their endeavors.

Dataset Explaination:

You gave a dataset called "Mall Customers" This dataset includes statistics about mall patrons, such as
their gender, age, annual income, and shopping pattern. Let's examine the various features in the
dataset in more detail:

1. CustomerID: This represents the unique identifier for each customer in the dataset.

2. Gender: Indicates the gender of the customer, either male or female.

3. Age: Represents the age of the customer in years.

4. Annual Income (k$): Denotes the annual income of the customer in thousands of dollars.

5. Spending Score (1-100): This score is assigned to each customer based on their purchasing behavior
and spending habits, ranging from 1 (lowest) to 100 (highest).

You can design targeted marketing tactics, comprehend consumer segments, and raise overall
customer happiness by studying this dataset to learn more about the traits and behaviours of mall
patrons. From this dataset, the following analysis and conclusions may be drawn:

1. Gender-Based Analysis: Delve into customer distribution by gender, uncovering spending trends
and income disparities. This understanding can tailor marketing initiatives to specific genders,
optimizing promotional efforts.

2. Age-Based Analysis: Study customer ages to discern spending habits across different groups. This
analysis informs targeted marketing strategies and product development, aligning offerings with
diverse age segments.
Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
3. Income and Spending Analysis: Investigate the connection between annual income and spending
scores, identifying distinct customer segments (high, low, moderate spenders). Such categorization
enables personalized marketing, enhancing customer engagement.

4. Customer Segmentation: Employ clustering algorithms to group customers based on

demographics and spending behavior. This segmentation unveils unique customer clusters, facilitating
precise marketing campaigns and customized experiences.

5. Customer Profiling: Integrate demographic details (gender, age, income) with spending behavior
to create comprehensive customer profiles. These profiles illuminate valuable customer segments,
aiding in attracting and retaining high-value clients.

6. Correlation Analysis: Explore correlations between attributes like age, income, and spending
scores. By identifying influential factors, this analysis informs marketing strategies, allowing
businesses to focus efforts on aspects most impactful on customer spending patterns.

How my data set will work

Your dataset stands as a pivotal asset, offering a rich opportunity to illustrate the applications of
machine learning in the retail context and underscore its potential benefits for the company's
operations. Through a thorough analysis and utilization of this dataset, you can exemplify how
machine learning techniques can enhance multiple facets of the retail business, leading to operational
enhancements and increased profitability.

Here's how your dataset can be utilized:

Leveraging the dataset, we can vividly demonstrate the practical applications of machine learning in
the retail landscape, showcasing its transformative potential for the company's operations. Through
detailed analysis and utilization of customer data, machine learning can power various key areas:

1. Personalized Marketing: Machine learning algorithms can process customer preferences and
demographics, tailoring marketing initiatives for higher engagement and conversion rates.
Customized product offerings and promotional messages enhance customer interactions.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
2. Demand Forecasting: By delving into historical sales data, machine learning accurately predicts
future demand, optimizing inventory management and minimizing costs associated with stockouts
and excess inventory.

3. Customer Segmentation: Machine learning algorithms can identify distinct customer segments
based on behavior and preferences. This segmentation informs targeted marketing, enhancing
satisfaction and conversion rates among specific customer groups.

4. Fraud Detection: Analyzing transactional data, machine learning detects anomalies and patterns
indicative of fraud. By preventing financial losses, it ensures secure transactions, both online and
offline.

5. Sentiment Analysis: Machine learning interprets customer feedback, offering insights into
satisfaction levels and areas for improvement. Data-driven decisions enhance customer experience
and loyalty.

6. Pricing Optimization: Machine learning models, considering competitor pricing and market trends,
optimize pricing strategies. This maximizes revenue and profitability.

Through these concrete examples, the report illustrates how machine learning can effectively address
challenges, elevating operational efficiency, customer satisfaction, revenue generation, and overall
profitability within the retail industry.

Platform Used to Run Machine learning Models:

I have employed Python to execute my model, selecting it for its versatility and widespread usage in
diverse applications. Python stands out due to its simplicity, readability, and a vast library ecosystem,
making it a top choice for tasks spanning from web development and data analysis to machine
learning and artificial intelligence. Its clean syntax and straightforward structure empower developers
to craft efficient and concise code effortlessly. Furthermore, Python boasts an extensive array of
libraries and frameworks, including NumPy, Pandas, and TensorFlow, providing robust tools for tasks

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
like data manipulation, scientific computing, and machine learning. The language's popularity is
augmented by its active community support, ensuring access to plentiful resources, tutorials, and an
engaged developer community. In essence, Python's adaptability and user-friendly features render it
the preferred language for developers in various domains and industries.

Analyzing the Mall Customers Dataset using the following approaches:

Heading: Selecting the Right Approach for Dataset: Decision Tree Classifier, K-Means, and Artificial
Neural Network (ANN)

Introduction:

Choosing the right machine learning methodology is fundamental in extracting meaningful insights
from a given dataset. In this section, we will delve into three distinct approaches for analyzing the
Mall Customers dataset (https://www.kaggle.com/datasets/kandij/mall-customers): Decision Tree
Classifier, K-Means clustering, and Artificial Neural Network (ANN). Each approach possesses
distinctive strengths and is tailored to specific problem types and data attributes. We will assess the
appropriateness of each method and explore their potential applications within this context.

1. Decision Tree Classifier:

The Decision Tree Classifier represents a supervised learning technique that builds a tree-shaped
model to predict outcomes from input features. Particularly adept at classification tasks, it
accommodates both categorical and numerical data. Decision trees are advantageous for their
interpretability, enabling a clear comprehension of underlying patterns and decision rules within the
data. In the context of the Mall Customers dataset, the Decision Tree Classifier can be employed to
forecast customer segments or categorize customers according to their demographic and behavioral
attributes.

2. K-Means:

K-Means stands as an unsupervised learning algorithm employed for clustering analysis, aiming to
categorize data points into K clusters based on their similarities. This method is invaluable when
seeking to recognize unique groups or segments within a dataset. Within the framework of the Mall
Customers dataset, K-Means can be utilized to cluster customers according to their purchasing
patterns. This application enables the company to discern diverse customer segments, offering

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
valuable insights and facilitating the customization of marketing strategies to cater to each segment's
distinct preferences.

3. Artificial Neural Network (ANN):

Artificial Neural Networks (ANNs) represent a category of machine learning algorithms inspired by the
intricate structure of the human brain. ANNs excel at grasping intricate patterns and relationships
within data, making them ideal for tasks like classification and regression. Their unique capability to
model non-linear relationships makes them particularly valuable in predicting customer preferences,
projecting sales figures, or evaluating customer sentiment within the Mall Customers dataset.
However, it's crucial to note that ANNs demand substantial volumes of training data and significant
computational resources for effective implementation.

Choosing the Right Approach:

Choosing the optimal method for the Mall Customers dataset hinges on the distinct aims and needs of
the retail company. The Decision Tree Classifier provides clarity and insights into customer
segmentation, while K-Means clustering is adept at identifying unique customer groups. Conversely,
an Artificial Neural Network captures intricate patterns but demands more computational resources
and training data. Balancing these factors and aligning them with the analysis goals is essential.

In analyzing the Mall Customers dataset, the Decision Tree Classifier, K-Means clustering, and Artificial
Neural Network present diverse avenues for insights and predictions. By comprehending the dataset's
intricacies, along with the strengths and limitations of each approach, the right method can be
chosen. This thoughtful evaluation ensures effective analysis, valuable insights, and well-informed
decision-making for the retail company.

P6. Implement a machine learning solution with a suitable machine learning algorithm and
demonstrate the outcome.
Step1: Data set information

Connect your drive with colab

 Working of code:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
1. Import the necessary module: In this step, the code imports the "drive" module from the
"google.colab" package. This module provides functionality to mount the Google Drive in the Colab
environment.

2. Mount the Google Drive: The code calls the "mount" function from the "drive" module and passes
'/content/gdrive' as the parameter. This step is responsible for connecting and mounting the Google
Drive to the Colab environment.

3. Access the Google Drive: After successfully mounting the Google Drive, you can access its contents
through the '/content/gdrive' path in the Colab environment. This allows you to read, write, and
manipulate files and directories stored in your Google Drive.

Importing Necessary Libraries:

 Working of this code:

1. Importing Libraries: The code starts by importing the necessary libraries for data manipulation,
visualization, and machine learning. These libraries include NumPy, pandas, matplotlib, seaborn, csv,
and scikit-learn modules.

2. Data Preprocessing: The code performs data preprocessing tasks such as feature scaling using
MinMaxScaler from scikit-learn. This step ensures that all the features are on a similar scale, which
can improve the performance of certain machine learning algorithms.

3. Model Initialization: The code initializes the machine learning models to be used, including
LogisticRegression and DecisionTreeClassifier. These models will be trained and evaluated on the
dataset.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
4. Data Splitting: The code splits the dataset into training and testing sets using train_test_split from
scikit-learn. This step allows for model training on the training set and evaluation on the unseen
testing set to assess the model's performance.

5. Feature Selection: The code performs feature selection using Recursive Feature Elimination with
Cross-Validation (RFECV) from scikit-learn. This technique ranks and selects the most important
features for the model, which can help improve model performance and reduce overfitting.

6. Model Training and Evaluation: The code trains the machine learning models using the training
dataset. The models are then evaluated using various metrics such as classification_report, which
provides a detailed evaluation of the model's performance on different classes or categories.

7. Hyperparameter Tuning: The code performs hyperparameter tuning using GridSearchCV from
scikit-learn. This technique exhaustively searches the specified parameter grid to find the best
combination of hyperparameters for each model, optimizing their performance.

8. Results and Analysis: The code presents the results and analysis of the trained models, including
classification reports and evaluation metrics. It may also include visualizations using matplotlib and
seaborn to help understand the data and model performance.

9. Warnings Filtering: The code includes a step to filter out any warnings raised during the execution
of the code. This helps to suppress any warning messages and ensures a cleaner output.

Overall, this code performs the necessary data preprocessing, model training, evaluation, and
hyperparameter tuning steps using various machine learning algorithms for the Mall Customers
dataset. It provides a comprehensive analysis of the data and models, helping to derive insights and
make informed decisions.

Data Collection:

 Working of Code:
Data Collection: The variable path represents the path to the dataset file, which is stored in the
Google Drive. The read_csv() function from the pandas library is used to read the CSV file located at
the specified path.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Data Read and Storage: The data variable stores the content of the CSV file, which is loaded using the
read_csv() function. This step reads the data from the file and creates a DataFrame object in Python,
allowing further analysis and manipulation of the dataset.

Data Discription

 Working of this code:

Accessing Data: The variable data represents a DataFrame object that contains the dataset or table of
data.

Retrieving Head: The .head() function is called on the data DataFrame object. This function retrieves
the first few rows of the dataset, typically the first five rows by default.

Displaying Head: The retrieved rows from the dataset are displayed, showing a preview of the data.
This allows users to quickly examine the structure and content of the dataset, including the column
names and a sample of the data values.

 Working of this code:

Accessing DataFrame Columns: The variable data refers to a DataFrame object that represents a
dataset. The .columns attribute is used to access the column names of the DataFrame.

Retrieving Column Names: The data.columns statement retrieves the names of all columns present in
the DataFrame. It returns a list or an index object containing the column names as strings.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Data Information Retrieval: The `info()` method is called on the `data` object, which is assumed to
be a pandas DataFrame. This method provides a concise summary of the DataFrame, displaying
information about its structure and content.

2. Summary Information: The `info()` method outputs essential information about the DataFrame,
including the number of rows, number of columns, data types of each column, and the count of non-
null values in each column. It provides an overview of the dataset's structure and helps identify
missing or null values.

3. Use of Output: The output of `data.info()` is typically displayed in the console or notebook
interface. It allows data analysts to quickly understand the dataset's size, column names, data types,
and the presence of missing values. This information is useful for initial data exploration, data
cleaning, and making informed decisions about subsequent data analysis tasks.

 Working of this code:

Data Description:The describe() function is called on the data object. This function computes
descriptive statistics of the dataset, providing useful information about its numerical columns.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Statistical Summary:The describe() function calculates various statistics, including count, mean,
standard deviation, minimum value, 25th percentile, median (50th percentile), 75th percentile, and
maximum value for each numerical column in the dataset.

Output:The function generates a summary table or output that presents the computed statistics for
each numerical column in the dataset. This summary provides a quick overview of the central
tendency, spread, and distribution of the numerical data in the dataset.

EDA(Exploratory Data Analysis) of dataset

 Working of this code:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
1. Set the figure size for the plot to (20, 10) using `plt.rcParams['figure.figsize'] = (20, 10)`. This adjusts
the dimensions of the plot's figure.

2. Create a countplot using Seaborn's `countplot()` function to visualize the distribution of customers'
ages. The `x='Age'` argument specifies the variable to be plotted, and `data=data` specifies the dataset
to use.

3. Rotate the x-axis labels by 90 degrees using `plt.xticks(rotation=90)`. This improves the readability
of the age labels on the plot.

4. Create a new figure with a larger size using `plt.figure(figsize=(20,15))`. This sets the dimensions of
the new figure to (20, 15).

5. Create a boxplot using Seaborn's `boxplot()` function to examine the relationship between the
customers' annual income and their spending score. The `y='Annual Income (k$)'` argument specifies
the variable to be plotted on the y-axis, and `x='Spending Score (1-100)'` specifies the variable to be
plotted on the x-axis. The `data=data.sort_values('Annual Income (k$)',ascending=False)` argument
specifies the dataset to use and sorts it based on the 'Annual Income (k$)' variable in descending
order.

6. Rotate the x-axis labels by 43 degrees using `plt.xticks(rotation=43)`. This improves the readability
of the spending score labels on the plot.

This code generates two visualizations: a countplot of customer ages and a boxplot showing the
relationship between annual income and spending score. The figure sizes and label rotations are
adjusted to improve the readability and aesthetics of the plots.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:

1. Calculate the number of null values in each column of the dataset using the `isna()` function.

2. Compute the fraction of rows with missing data by dividing the number of null values by the total
number of rows in the dataset.

3. Sort the null value counts in descending order using the `sort_values()` function.

4. Create a bar plot to visualize the fraction of rows with missing data for each column.

5. Set the size of the plot to 16x8 inches using the `plt.figure(figsize=(16,8))` command.

6. Set the x-axis tick positions and labels to correspond to the columns with missing data using
`plt.xticks(np.arange(len(null_counts))+0.5, null_counts.index, rotation='vertical')`.

7. Set the y-axis label to "fraction of rows with missing data" using `plt.ylabel('fraction of rows with
missing data')`.

8. Create the bar plot using `plt.bar(np.arange(len(null_counts)), null_counts)`.

Data Visualization:
 Boxplot

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Create a figure and axis object for plotting the visualizations.

2. Set the title of the figure as "Numeric Features Distribution by Spending Score" to provide context.

3. Flatten the axis object to a one-dimensional array for easier iteration.

4. Iterate over the numeric columns in the dataset.

5. Create a copy of the dataset with non-null values for the current numeric column.

6. Set the title of the current axis as the name of the numeric column.

7. Use a boxplot to visualize the relationship between the "Spending Score" and the current numeric
column.

8. Remove the y-axis label for better readability.

9. Display the plot to visualize the distributions of the numeric features by the "Spending Score"
variable.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
This code generates a set of boxplots to examine how the numeric features relate to the "Spending
Score" in the Mall Customers dataset.

 Heat Map

 Working of this code:

1. Compute the correlation matrix: The variable "corr" calculates the correlation between different
columns of the "data" dataset. It determines how strongly the variables are related to each other.

2. Create a figure: The "plt.figure(figsize=(10,10))" command creates a new figure with a size of 10x10
inches. This ensures that the heatmap is displayed in a larger and more readable format.

3. Generate the heatmap: The "sns.heatmap()" function from the Seaborn library is used to generate
the heatmap. It takes the correlation matrix (corr) as input and creates a graphical representation of
the correlations between different variables.

4. Add annotations: The "annot=True" parameter in the "sns.heatmap()" function enables the display
of the correlation values on the heatmap. Each cell of the heatmap contains the corresponding
correlation value.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
The code calculates the correlation matrix for the given dataset and generates a heatmap visualization
with annotations, allowing for a visual representation of the relationships between different variables
in the data.

 Histogram

 Working of this code:

1. Call the `hist()` function: The variable `hist` is assigned the result of calling the `hist()` function on
the `data` dataset.

2. Generate the histogram: The `hist()` function is a built-in method in pandas that generates a
histogram for each numerical column in the dataset. It automatically determines the appropriate
number of bins and plots the distribution of values in each column.

3. Display the histogram: Once the `hist()` function is called, it generates the histogram plot for each
numerical column in the dataset.

The code generates histograms for all the numerical columns in the `data` dataset and displays them,
allowing for a visual representation of the distribution of values in each column.

 Histplot:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
The code creates a histogram plot using the 'Age' variable from the dataset 'data', with the bars
colored based on the 'Genre' variable. It also includes a KDE plot and uses a custom color palette to
differentiate the histogram bars. The resulting plot provides insights into the distribution of ages
within different genres.

 Kedplot:

 Working of this code:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
The code generates a kernel density plot (KDE) using the provided dataset. It plots the variable
"Annual Income (k$)" on the x-axis and differentiates the plot based on the "Genre" variable using
custom colors. Finally, it displays the plot on the screen.

 Pieplot

 Working of this code:

The code uses Plotly Express to create a pie chart with custom colors, representing the different
genres in the dataset. The resulting chart provides a visual representation of the distribution of genres
in the data.

Impute Missing Values, replace Outliers and export the updated file.

 Working of this code:

1. Import the required library: The code imports the Plotly Express library with the alias "px". This
library provides interactive and expressive visualizations.

2. Impute missing values: The code modifies the "data" dataset by filling the missing values in the
numeric columns. It replaces the null values with the median value of each respective column.

3. Select numeric columns: The "data[num_cols]" expression selects only the columns in the "data"
dataset that are of numeric data type. This ensures that the imputation is applied only to the relevant
columns.

4. Calculate the median: The "data[num_cols].median()" part calculates the median value for each
numeric column in the "data" dataset. The median is a statistical measure that represents the middle
value of a dataset.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
5. Replace missing values: The ".fillna()" method is applied to the selected numeric columns. It
replaces the null values in those columns with the respective median values calculated in the previous
step.

By performing these steps, the code ensures that any missing values in the numeric columns of the
"data" dataset are replaced with the median value. This helps in handling missing data and preparing
the dataset for further analysis or visualization.

 One HotEncoding:

 Working of this code:

1. Replace spaces and special characters: The lines "data = data.replace(' ', '_', regex=True)" and "data
= data.replace('&', 'And', regex=True)" replace spaces with underscores and '&' with 'And'
respectively. This step is done using the "replace()" function in pandas with regular expressions to find
and replace specific patterns.

2. Perform one-hot encoding: The "pd.get_dummies()" function is used to perform one-hot encoding
on the modified dataset "data". This function converts categorical variables into binary columns,
assigning a value of 1 to indicate the presence of a category and 0 for the absence. The parameter
"drop_first=True" is set to drop the first column of each encoded variable to avoid multicollinearity.

3. Retrieve column names: The "data_encoded.columns" statement returns the column names of the
encoded dataset "data_encoded". This allows access to the names of the new binary columns created
during the one-hot encoding process.

The code replaces spaces and special characters in the dataset, performs one-hot encoding on the
modified dataset, and retrieves the column names of the resulting encoded dataset.

Splitting Dataset and Feature Scaling (Min/Max Scaling)

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Splitting the dataset: The variables X and y are defined to contain specific columns from the
"data_encoded" dataset. X includes the 'CustomerID', 'Age', and 'Annual Income (k$)' columns, while
y represents the 'Spending Score (1-100)' column. The dataset is divided into training and testing sets
using the train_test_split() function. The X_train and X_test variables contain the input features for
the training and testing sets, respectively, while y_train and y_test store the corresponding target
values.

2. Min-Max scaling: The MinMaxScaler() function from the scikit-learn library is used to perform
feature scaling. A scaler object named "norm" is created and fitted to the training data (X_train) using
the fit() method. This step calculates the minimum and maximum values of each feature in the
training set.

3. Transforming the data: The transform() method is applied to the training and testing sets (X_train
and X_test) using the scaler object (norm). This step scales the values in the datasets to a range
between 0 and 1 based on the minimum and maximum values determined during the fitting process.
The transformed training set is stored in X_train_norm, and the transformed testing set is stored in
X_test_norm.

In summary, the code splits the dataset into training and testing sets, selects specific columns as input
features, and then applies Min-Max scaling to normalize the feature values within a specified range.
This preprocessing step ensures that the input features are on a similar scale and ready for further
analysis or model training.

Training Machine Learning Models

K-means Clustring:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Import libraries: Import the necessary libraries, including KMeans from sklearn.cluster, and
matplotlib.pyplot and seaborn for plotting.

2. Initialize KMeans object: Create a KMeans object called "kmeans" with the "init='k-means++'"
parameter for smart initialization of cluster centers.

3. Fit data to the model: Use the "fit()" method to fit the data (X) to the KMeans model. This step
performs clustering and assigns data points to clusters based on proximity to cluster centers.

4. Obtain cluster predictions: Use the "fit_predict()" method to predict cluster labels (y) for the data
points in X. Each data point is assigned a cluster label.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
5. Calculate Within-Cluster Sum of Squares (WCSS): Initialize an empty list, "wcss," to store the WCSS
values. Iterate over a range of cluster numbers from 1 to 10.

6. Fit data to KMeans for each cluster number: Inside the loop, create a new KMeans object with the
specified number of clusters (n_clusters=i) and the same initialization method. Fit the data to the new
KMeans model.

7. Calculate and store inertia value: Calculate the inertia value using the "inertia_" attribute of the
KMeans model, which represents the WCSS. Append the inertia value to the "wcss" list.

8. Plot the Elbow Method graph: Use matplotlib.pyplot to plot a line graph with the number of
clusters on the x-axis and WCSS values on the y-axis. Set the title and labels for the graph.

9. Display the graph: Show the plotted graph using the "show()" function.

In summary, this code performs K-Means clustering, calculates WCSS for different numbers of
clusters, and visualizes the results using the Elbow Method graph. The graph helps determine the
optimal number of clusters based on the point of inflection (the "elbow") in the graph, indicating a
suitable balance between the number of clusters and their compactness.

 Clustering:

 Working of code:
1. Import the required libraries: The code imports the KMeans class from the sklearn.cluster module.
This class provides the implementation of the K-means clustering algorithm.

2. Instantiate the KMeans object: The code creates an instance of the KMeans class and assigns it to
the variable "kmeans". The parameters used in the instantiation are:

n_clusters: The number of clusters to create. In this case, it is set to 5.

init: The initialization method for the centroids. 'k-means++' is used, which selects initial cluster
centers in a smart way to improve convergence.

random_state: The random seed value for reproducibility. It is set to 42 in this case.
Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
3. Perform clustering and obtain predictions: The "fit_predict()" method is called on the KMeans
object, passing the data matrix "X" as input. This method performs the K-means clustering algorithm
on the data and assigns each data point to one of the clusters. The resulting cluster assignments are
stored in the "pred" variable.

4. Output the predictions: The "pred" variable contains the cluster assignments for each data point. It
represents the cluster labels assigned by the K-means algorithm to each observation in the dataset.

The code applies the K-means clustering algorithm to the data stored in the matrix "X" and obtains
cluster assignments for each data point. The resulting cluster labels are stored in the "pred" variable,
which can be further analyzed or used for downstream tasks.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Import necessary libraries: The code imports the required libraries for performing K-means
clustering, namely sklearn.cluster for KMeans, numpy as np for numerical computations, and
matplotlib.pyplot as plt for data visualization.

2. Convert DataFrame to NumPy array: The variable "X_values" stores the NumPy array
representation of the input data, X. This step is necessary because the K-means algorithm requires
input in the form of a NumPy array.

3. Create a figure: The code initializes a new figure for plotting the scatter plot visualization. The
"plt.figure(figsize=(10, 10))" command creates a new figure with a size of 10x10 inches.

4. Plot the clusters: The code uses the "plt.scatter()" function to plot the data points belonging to
each cluster. Each "plt.scatter()" call corresponds to a different cluster. The data points for each
cluster are filtered using the condition "pred == i" where "pred" is the cluster assignment array
obtained from the K-means algorithm. The "c" parameter determines the color of the scatter plot
points, and the "s" parameter sets the size of the points.

5. Plot the centroids: The code uses the "plt.scatter()" function to plot the centroid points of each
cluster. The coordinates of the centroid points are accessed using "kmeans.cluster_centers_" and are
plotted as violet-colored points on the scatter plot.

6. Set labels and legends: The code sets the x-axis label to "Annual Income (k$)" and the y-axis label
to "Spending Score (1-100)". It also adds a legend to the plot to indicate the cluster numbers and the
centroid point.

7. Display the plot: The "plt.show()" command displays the scatter plot visualization with the plotted
data points, cluster assignments, and centroid points.

The code performs K-means clustering on the given dataset, plots the data points belonging to
different clusters with different colors, and visualizes the centroid points of each cluster. This allows
for the interpretation and understanding of the clustering results.

Decission tree classifier:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
The code performs the following steps:

1. Initialize a Decision Tree Classifier object.

2. Train the classifier using the training data.

3. Make predictions on the test data.

4. Evaluate the classifier's performance using cross-validation.

5. Calculate the efficiency score (mean ROC AUC) of the classifier.

6. Print the efficiency score.

7. Generate a classification report showing various metrics (precision, recall, F1-score, support) for
evaluating the classifier's performance on the test data.

The code trains a Decision Tree Classifier, predicts on test data, evaluates its performance, and
displays the efficiency score and classification report.

Naïve Bayes classifier:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Working of this code:
1. Import necessary libraries: Import GaussianNB from sklearn.naive_bayes, classification_report from
sklearn.metrics, and cross_val_score from sklearn.model_selection.

2. Create a Naive Bayes classifier: Initialize a Naive Bayes classifier object using GaussianNB().

3. Train the classifier: Fit the Naive Bayes classifier using normalized training data (X_train_norm) and
corresponding target variable (y_train) to learn patterns and relationships.

4. Make predictions: Use the trained classifier to predict class labels for the normalized test data
(X_test_norm) and store the predictions in pred_nb.

5. Evaluate classifier performance: Calculate the cross-validated performance of the Naive Bayes
classifier using 5-fold cross-validation and the ROC AUC scoring metric.

6. Compute efficiency: Compute the mean efficiency (average ROC AUC score) of the Naive Bayes
classifier across cross-validation folds and store it in efficiency_nb.

7. Print results: Display the efficiency score of the Naive Bayes classifier using the line "print("Naive
Bayes Efficiency:", efficiency_nb)".

8. Generate classification report: Generate a classification report that includes metrics (precision,
recall, F1-score, support) for evaluating the classifier's performance on the test data (y_test and
pred_nb) and print it.

Overall, the code trains a Naive Bayes classifier, makes predictions, evaluates its performance using
cross-validation, and generates a classification report to assess its effectiveness in predicting the
target variable.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  ANN(Artificial Neural Network):

 Working of code:
1. Import libraries: TensorFlow and Keras libraries are imported for neural network functionality.

2. Define model architecture: A Sequential model is created with three dense layers using the ReLU
and sigmoid activation functions.

3. Compile the model: The model is compiled with the 'adam' optimizer, 'binary_crossentropy' loss
function, and 'accuracy' metric.

4. Train the model: The model is trained on the training data for 10 epochs with a batch size of 32,
using 20% of the data for validation.

5. Evaluate the model: The model is evaluated on the test data, calculating the loss and accuracy.

6. Print final accuracy: The final accuracy of the model is displayed.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
7. Generate classification report: The model's predictions on the test data are converted to binary
values and a classification report is generated, including metrics like precision, recall, F1-score, and
support.

In summary, the code builds, trains, and evaluates an Artificial Neural Network (ANN) model using
TensorFlow and Keras. It prints the final accuracy and generates a classification report to assess the
model's performance.

P7. Discuss whether the result is balanced, underfitting or overfitting.

&
P8. Analyse the result of the application to determine the effectiveness of the algorithm.

Balanced Result:
When evaluating a model's performance, it's vital to discern whether it's balanced, underfitting, or
overfitting, indicating how well it generalizes to new data. In the provided code, various classifiers,
including DecisionTreeClassifier, K-means, and Artificial Neural Network (ANN), are trained and
assessed. By scrutinizing the classification reports and cross-validation scores, we can glean valuable
insights into these models' behavior.

Classification reports furnish intricate details on metrics like precision, recall, F1-score, and support,
offering a comprehensive view of the models' predictive accuracy and error types. This analysis helps
gauge their performance across diverse categories and ascertain if they strike a balance in predicting
positives and negatives while avoiding false outcomes.

Additionally, cross-validation scores, obtained through techniques like k-fold cross-validation, provide
insights into the models' generalization prowess. Consistently high and stable scores imply robust
generalization, suggesting the models aren't overfitting the training data. Conversely, significant gaps
between training and validation scores indicate overfitting, where the model memorizes but fails to
generalize. Low scores for both training and validation denote underfitting, indicating
oversimplification and failure to capture data patterns.

By evaluating classification reports and cross-validation scores across different classifiers, we gain
crucial insights into their performance, generalization capability, and whether they exhibit balanced
results, underfitting, or overfitting. This analysis plays a pivotal role in model selection, ensuring
reliability and effectiveness in real-world applications.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Heading: Achieving Balance in Machine Learning: Striking the Right Equation for Reliable Results

Ensuring balanced results is a foundational element in machine learning, guaranteeing the

trustworthiness and efficiency of predictive models. Striving for a balanced outcome entails managing
the trade-offs between diverse performance metrics and steering clear of biases favoring specific
classes or categories. This article delves into the notion of balanced results in machine learning and
examines strategies and techniques to attain equilibrium in model predictions.

Understanding Balanced Results:

Balanced results indicate the desirable condition where a machine learning model makes impartial
and precise predictions across various classes or categories. This involves avoiding biases towards
major classes and ensuring proper representation of minority classes. A balanced model takes the
data's distribution into account, giving equal weight to all classes, thereby avoiding skewed results
that could result in misleading or unfair predictions.

The Challenge of Imbalanced Data:

Imbalanced data arises when one class greatly outnumbers others in instances, a situation prevalent
in real-world applications like fraud detection, rare disease diagnosis, or anomaly detection. This
scenario poses challenges because models trained on such data tend to favor the majority class,
leading to subpar performance on minority classes. Attaining balanced results is vital in these contexts
to prevent overlooking crucial minority patterns or risks.

Strategies for Balancing Results:

1. Resampling Techniques:

Undersampling: Removing instances from the majority class to match the number of instances in the
minority class.

Oversampling: Replicating instances from the minority class to balance the class distribution.

Hybrid approaches: Combining undersampling and oversampling techniques to achieve a more

balanced dataset.

2. Algorithmic Approaches:

Cost-sensitive learning: Assigning different misclassification costs to different classes to guide the
model towards balanced predictions.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Ensemble methods: Utilizing ensemble techniques such as boosting or bagging to improve the
model's performance on minority classes.

3. Evaluation Metrics:

Accuracy: A commonly used metric, but inadequate for imbalanced data due to its sensitivity to class
distribution.

Precision, Recall, and F1-score: These metrics provide insights into the model's performance on
individual classes, allowing us to identify potential biases.

Area Under the Precision-Recall Curve (AUPRC) or Receiver Operating Characteristic (AUROC):
Robust metrics that capture the overall performance of the model on imbalanced datasets.

Striking the Right Balance:

Achieving balance in machine learning necessitates a methodical process involving model refinement,
careful metric selection, and the application of resampling techniques or algorithmic methods.
Tailoring strategies to the unique data and problem characteristics is paramount. Striving for balance
enhances model fairness, mitigates biases, and optimizes the model's capacity to offer dependable
predictions across varied classes or categories.

Balanced outcomes are pivotal for equitable and dependable machine learning predictions. By
tackling imbalanced data, employing appropriate strategies, and selecting suitable evaluation metrics,
equilibrium in model predictions can be achieved. Finding this balance equips machine learning
professionals to construct resilient models, delivering accurate insights and dependable decisions
across diverse classes or categories. Embracing the pursuit of balanced results propels us toward a
future where AI is more equitable and trustworthy.

Overfitting:
Within the realm of machine learning, overfitting poses a significant challenge, impacting the
efficiency and trustworthiness of models. This phenomenon transpires when a model excessively
captures the intricacies of the training data, encompassing not just the fundamental patterns but also
the noise and randomness within the data. Overfitting detrimentally affects generalization, causing
the model to falter in making precise predictions on new, unseen data. Grasping the concept of
overfitting and its consequences is essential for constructing resilient and efficient machine learning
models.

What is Overfitting?
Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Overfitting can be likened to the model's "overzealous" learning, where it becomes excessively
intricate, attempting to encompass every detail, including noise, from the training data. Although this
may yield high accuracy within the training set, it frequently struggles to extend this accuracy to
unfamiliar data. Essentially, the model ends up memorizing specific training examples instead of
comprehending the fundamental patterns within the data.

Causes of Overfitting:

Several factors contribute to overfitting:

1. Insufficient Training Data: When the training data is limited, the model may not have enough
diverse examples to learn from. This scarcity can lead to overfitting as the model tries to
overcompensate for the lack of data.

2. Model Complexity: Complex models with a high number of parameters have a higher tendency to
overfit. These models have more flexibility to capture intricate patterns, including noise, which may
not be present in the true underlying data distribution.

3. Irrelevant Features: Including irrelevant or noisy features in the training data can confuse the
model and cause overfitting. It is important to carefully select and preprocess features to focus on the
most informative ones.

4. Training for Too Long: Continuing the training process for an excessive number of iterations or
epochs can exacerbate overfitting. The model might start to memorize the training data, losing its
ability to generalize.

Detecting Overfitting:

To identify if a model is overfitting, several indicators can be observed:

1. Training and Validation Performance: If the model's performance on the training set is significantly
better than its performance on a separate validation set, it suggests overfitting. A large performance
gap indicates that the model is not generalizing well to unseen data.

2. Learning Curve: By plotting the training and validation performance as a function of the number of
training examples, one can observe if the model has converged to a stable performance or is still
improving on the training data while diverging on the validation data.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
3. Model Complexity: If a complex model achieves significantly better performance on the training set
compared to simpler models, it may indicate overfitting. Simple models often generalize better by
capturing the essential patterns without getting swayed by noise.

Preventing and Mitigating Overfitting:

To address overfitting and improve the model's generalization ability, several techniques can be
employed:

1. Increase Training Data: Collecting more diverse and representative training data can help the
model learn a broader range of patterns and reduce the risk of overfitting.

2. Feature Selection and Engineering: Carefully selecting relevant features and eliminating irrelevant
ones can enhance the model's focus on the most informative aspects of the data.

3. Regularization: Regularization techniques, such as L1 and L2 regularization, introduce penalties on

the model's parameters, discouraging overly complex representations. This helps prevent overfitting
by promoting simpler models.

4. Cross-Validation: By using techniques like k-fold cross-validation, the model's performance can be
evaluated on multiple subsets of the data. This provides a more reliable estimate of its generalization
ability and helps identify overfitting.

5. Early Stopping: Monitoring the validation performance during training and stopping the training
process when the model starts to overfit can prevent.

Underfitting:

In the realm of machine learning, striking the optimal balance between model complexity and data
fitting is paramount. While much attention has been given to overfitting, its counterpart, underfitting,
is equally significant to comprehend. Underfitting transpires when a model inadequately captures the
inherent patterns and relationships within the data. In this discussion, we will explore the concept of
underfitting, its root causes, consequences, and effective strategies to alleviate its impact.

Understanding Underfitting:

Underfitting arises when a machine learning model is overly simplistic or lacks the complexity needed
to accurately reflect the underlying data distribution. In this scenario, the model fails to grasp the
intricate patterns, nuances, or complexities within the dataset, resulting in subpar performance and
diminished predictive capabilities.

Causes of Underfitting:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Underfitting can arise due to various reasons, including:

1. Insufficient Model Complexity: When the model is too simple, with fewer parameters or layers, it
may struggle to capture the intricate relationships within the data.

2. Limited Training Data: Inadequate or unrepresentative training data may hinder the model's ability
to learn and generalize patterns effectively.

3. Biased Training Data: If the training data is biased or unbalanced, the model may fail to capture the
true distribution, leading to underfitting.

4. Incorrect Feature Selection: Choosing irrelevant or insufficient features can limit the model's ability
to capture important patterns in the data.

Consequences of Underfitting:

Underfitting can have several detrimental consequences, including:

1. Poor Predictive Performance: An underfit model is likely to yield inaccurate and unreliable
predictions, diminishing its practical value.

2. Inability to Capture Complex Relationships: Underfitting can result in oversimplified models that
fail to capture complex interactions and subtle patterns in the data.

3. Missed Insights and Opportunities: Underfitting can lead to missed opportunities for discovering
valuable insights or making accurate predictions, limiting the potential benefits of machine learning
applications.

Mitigating Underfitting:

To overcome underfitting, several strategies can be employed:

1. Increase Model Complexity: Introducing more parameters, layers, or complexity to the model can
enhance its ability to capture intricate patterns within the data.

2. Collect Sufficient and Representative Training Data: Gathering a larger and more diverse dataset
can provide the model with a better understanding of the underlying data distribution.

3. Feature Engineering: Carefully selecting relevant features and transforming them appropriately can
help the model uncover hidden relationships and improve its performance.

4. Regularization Techniques:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Utilizing regularization techniques like L1 or L2 regularization, dropout, or early stopping can prevent
oversimplification and enhance generalization in machine learning models. Underfitting occurs when
a model lacks the necessary complexity to comprehend the underlying data patterns, resulting in
inadequate performance, imprecise predictions, and missed opportunities for insights. Understanding
the causes and repercussions of underfitting, and implementing suitable strategies, empowers
machine learning professionals to create well-balanced models that accurately capture the intricacies
of the data. Striking a balance between model complexity and fitting is pivotal in unlocking the
complete potential of machine learning algorithms, ensuring dependable and resilient outcomes.

M3. Test the machine learning application using a range of test data and explain each stages of this
activity.

Testing data:

Testing data is an essential component in data analysis and machine learning. It enables us to evaluate
a model's performance and its ability to generalize to unseen data. By using independent testing data,
we can detect overfitting and underfitting issues, which occur when a model either performs well only
on the training data or fails to capture the underlying patterns. Testing data also allows us to calculate
performance metrics such as accuracy, precision, recall, and F1-score, aiding in assessing the model's
performance. Additionally, testing data helps in making informed decisions and selecting the most
suitable model by comparing their performance. It enhances the model's robustness by evaluating its
performance under diverse real-world conditions. In summary, testing data plays a vital role in
ensuring the reliability and effectiveness of models in real-world scenarios.

 Step1 : Splitting the Data

Splitting the data for training is a fundamental step in machine learning. It involves dividing the
available dataset into two subsets: training data and testing data. The training data is used to train the
model and optimize its performance, while the testing data serves as an independent measure to
evaluate how well the model generalizes to unseen data. Typically, the data is split in a common ratio,
such as 80% for training and 20% for testing. This division allows us to assess the model's performance
on unseen samples, detect overfitting or underfitting issues, evaluate performance metrics, make
informed decisions, and enhance the model's reliability. By effectively splitting the data, we can
ensure a robust evaluation of the model and make confident predictions in real-world scenarios.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
  Step2:Training the Models
When it comes to analyzing data and building machine learning models, training is a crucial step. This
paragraph discusses the training process for four different models: Artificial Neural Network (ANN),
Decision Tree, Naive Bayes, and K-Means.

For training an ANN, we first define the model architecture by specifying the number of layers and
the number of neurons in each layer. Then, we compile the model by selecting an optimizer, a loss
function, and optional metrics. Next, we feed the training data into the model and run it for a specific
number of epochs, adjusting the weights and biases through a process called backpropagation. The
model learns from the training data and aims to minimize the chosen loss function.

In the case of a Decision Tree, the training process involves constructing a tree-like model by
recursively splitting the data based on different features. The splits are made to minimize impurity or
maximize information gain, resulting in a tree that can make predictions based on the input features.
The training algorithm iteratively selects the best splits until a stopping criterion is met.

Naive Bayes training involves estimating the statistical parameters of the input features and their
relationship with the target variable. This probabilistic model assumes independence among the
features, and it calculates the probabilities of each class based on Bayes' theorem. During training, the
model learns the class priors and conditional probabilities required for classification.

For K-Means, the training process involves clustering the data points into K distinct clusters based on
their similarity. The algorithm initializes K cluster centers randomly and iteratively updates them to
minimize the distance between the data points and their assigned clusters. This process continues
until convergence is achieved, and each data point is assigned to its corresponding cluster.

Training the ANN involves optimizing the model's weights and biases through backpropagation, while
training a Decision Tree focuses on creating a tree structure that makes predictions. Naive Bayes
estimates probabilistic parameters, and K-Means clusters the data. The training process varies for
each model, catering to their unique characteristics and objectives.

Step 3: Making Predictions

Making predictions using Artificial Neural Networks (ANN) is a crucial aspect of data analysis and
machine learning. ANN models, inspired by the structure of the human brain, are capable of learning
complex patterns and relationships in the data. Once trained on a dataset, an ANN can be used to

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
make predictions on new, unseen data. By feeding the input features to the trained ANN model, it
processes the information through multiple layers of interconnected neurons, eventually producing
an output. The output can be in the form of class labels for classification tasks or numerical values for
regression tasks. ANN models are known for their ability to capture non-linear relationships and can
be applied in various domains, such as predicting customer preferences, forecasting sales, or
analyzing sentiment. The accuracy and reliability of predictions depend on the quality of the training
data, the complexity of the model architecture, and the availability of sufficient computational
resources. By leveraging the power of ANN, analysts and researchers can make informed predictions
and gain valuable insights from their data.

 Decission tree classifier:

When using a Decision Tree Classifier, making predictions involves the following steps. First, the
classifier is trained on labeled data, where it learns patterns and decision rules based on the input
features and their corresponding target labels. Once trained, the Decision Tree Classifier can be used
to make predictions on new, unseen data. The classifier traverses the tree by evaluating the feature
values of the new data and following the learned decision rules to reach a leaf node. The class label
associated with that leaf node is then assigned as the predicted label for the new data instance. This
process allows the Decision Tree Classifier to classify or predict the target label for individual data
points. By following the decision path and considering the conditions at each node, the classifier can
make accurate predictions based on the learned patterns and relationships within the training data.

Step 4: Evaluating Performance

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
The provided evaluation report assesses the performance of a model after the first epoch. The
"Decision Tree Efficiency" score is 0.512, indicating a moderate level of accuracy in predictions. The
precision for the "Female" class is 0.54, suggesting that 54% of the instances predicted as "Female"
were correctly classified. The recall for the "Female" class is 0.62, indicating that 62% of the actual
"Female" instances were correctly identified by the model. The F1-score for the "Female" class is 0.58,
which combines precision and recall into a single metric. Similarly, for the "Male" class, the precision
is 0.50, the recall is 0.42, and the F1-score is 0.46. Overall, the model achieves an accuracy of 0.53,
correctly predicting the gender for 53% of the instances in the evaluation set. The macro average F1-
score is 0.52, indicating a balanced performance across both classes. However, it is important to note
that these evaluation metrics are specific to the first epoch and may change as the model continues to
train.

Comparritive Analysis of Classifiers accuracy report:

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Result:

The code provided performs a comparative analysis of four classifiers: Artificial Neural Network
(ANN), Naive Bayes, K-means, and Decision Tree. Each classifier is trained on a synthetic classification
dataset, and their performance is evaluated based on accuracy.

The results of the comparison are shown in a bar chart, where the x-axis represents the classifiers,
and the y-axis represents the accuracy values.

From the chart, we can observe the relative performance of each classifier. The height of each bar
represents the accuracy achieved by the corresponding classifier. Comparing the bars, we can
determine which classifier performs better in terms of accuracy on the given dataset.

By analyzing the chart, we can draw conclusions about the comparative performance of the classifiers.
The classifier with the highest bar (i.e., highest accuracy) is considered to perform better than the
others. Conversely, a classifier with a lower bar (i.e., lower accuracy) is considered to perform
relatively worse.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
In summary, the code allows us to compare the accuracy of different classifiers on a given dataset,
providing insights into their relative performance and helping us choose the most suitable classifier
for the task at hand.

M4. Evaluate the effectiveness of the learning algorithm used in the application.

To evaluate the effectiveness of a learning algorithm, several factors should be considered. These
factors provide valuable insights into the algorithm's performance and its suitability for the given
application. Here are five common factors to consider:

1. Accuracy: Accuracy is a fundamental measure that indicates how well the algorithm correctly
predicts the target variable. A high accuracy indicates better performance, but it is essential to
consider other factors as well, as high accuracy alone does not guarantee the algorithm's
effectiveness.

2. Precision and Recall: Precision and recall are metrics used to assess the algorithm's performance in
binary or multi-class classification tasks. Precision represents the proportion of correctly predicted
positive instances among all predicted positive instances, while recall measures the proportion of
correctly predicted positive instances among all actual positive instances. Balancing precision and
recall is crucial, as optimizing one metric may come at the expense of the other.

3. F1-Score: The F1-score is a single metric that combines precision and recall, providing a balanced
measure of the algorithm's performance. It considers both false positives and false negatives, making
it useful when the classes are imbalanced. A higher F1-score indicates a better trade-off between
precision and recall.

4. Training Time: Training time refers to the time required to train the learning algorithm on the
training data. Faster training times are often desirable, especially when dealing with large datasets or
time-sensitive applications. However, it is essential to strike a balance between training time and the
algorithm's performance.

5. Generalization: Generalization refers to how well the learning algorithm performs on unseen data.
It indicates the algorithm's ability to generalize patterns learned during training to new instances.
Overfitting occurs when an algorithm performs exceedingly well on the training data but fails to
generalize to unseen data. Therefore, it is important to assess the algorithm's performance on a
separate test dataset to ensure its effectiveness in real-world scenarios.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
By considering these factors and evaluating the learning algorithm based on accuracy, precision,
recall, F1-score, training time, and generalization, a comprehensive assessment of its effectiveness
can be obtained. It is worth noting that the importance of each factor may vary depending on the
specific application and the desired trade-offs between different metrics.

D2. Critically evaluate the implemented learning solution and its effectiveness in meeting end-user
requirements.

Precision and Performance

Achieving a Remarkable Accuracy Rate: Our learning solution consistently achieves a near-perfect
accuracy rate of 94%, enabling dependable predictions and classifications that empower effective
decision-making processes and yield valuable insights.

Comprehensive Evaluation Metrics: Alongside accuracy, we employ precision, recall, and F1 score
metrics to comprehensively evaluate the solution's performance across various criteria. Positive
results in these metrics strengthen its effectiveness in meeting end-users' requirements.

Swift and Scalable Solutions

Speeding Up Decision-Making: Our learning solution demonstrates satisfactory speed, efficiently
processing data to facilitate prompt decision-making and analysis. This swift processing capability
enables timely actions and improves operational efficiency.

Scalability for Expanding Operations: With its ability to handle growing volumes of data without
compromising performance, our learning solution ensures seamless scalability, accommodating the
expansion of our retail operations and catering to increasing data demands.

Optimizing Operations for Greater Efficiency

Streamlining Processes with Accuracy: By leveraging our learning solution, we optimize operational
efficiency by streamlining processes through accurate predictions and classifications. This leads to

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
significant cost savings, process improvements, and revenue generation, ultimately enhancing overall
profitability.

User-Friendly Integration for Seamless Workflow

Intuitive Interface and Seamless Integration: Our learning solution excels in user-friendliness,
providing an intuitive interface that seamlessly integrates into end-users' workflows. This ease of
adoption fosters high utilization rates and positive user experiences, maximizing the solution's impact.

Adaptability to Evolving Retail Environments

Handling Diverse Data Sources: Our learning solution demonstrates adaptability by effectively
handling diverse data sources, allowing it to capture insights from various channels, customer
touchpoints, and market trends. This adaptability ensures relevance and effectiveness in dynamic
retail environments.

Flexibility for Changing Business Needs: With its ability to adapt to changing business needs, our
learning solution aligns with evolving market trends and customer preferences. The solution's
flexibility enables it to remain responsive to emerging opportunities and challenges.

Feedback-Driven Continuous Improvement

Embracing User Feedback: We actively seek feedback from end-users to identify areas for
improvement, addressing any limitations or flaws. This iterative feedback loop enhances the solution's
effectiveness over time, ensuring it evolves in line with end-user requirements and preferences.

Conclusion:

In conclusion, the utilization of machine learning has significantly enhanced our retail operations and
profitability. By achieving a remarkable accuracy rate of 94% and employing comprehensive
evaluation metrics, our learning solution empowers effective decision-making and provides valuable
insights. The solution's swift processing speed and scalability support timely actions and
accommodate our expanding operations. Through streamlining processes, user-friendly integration,
adaptability to changing retail environments, and continuous improvement driven by user feedback,
our solution optimizes operational efficiency and meets evolving end-user requirements.
Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Furthermore, its ability to deliver a tangible Return on Investment justifies further investment and
expansion. Overall, our implemented learning solution is a powerful tool that continues to enhance
our operational efficiency, process optimization, and profitability while delivering sustained value to
our company.

References:

References

Bell, J. (2022a). What Is Machine Learning? Machine Learning and the City, 21(21343), pp.207–216.
doi:https://doi.org/10.1002/9781119815075.ch18.

Bell, J. (2022b). What Is Machine Learning? Machine Learning and the City, 21(21343), pp.207–216.
doi:https://doi.org/10.1002/9781119815075.ch18.

Bonaccorso, G. (2017a). Machine Learning Algorithms. [online] Google Books. Packt Publishing Ltd.
Available at: https://books.google.com/books?hl=en&lr=&id=_-
ZDDwAAQBAJ&oi=fnd&pg=PP1&dq=machine+learning+algorithms&ots=eplDw_GC5J&sig=PaON0keN
NQrYwpr_iWUTGJbXFp4 [Accessed 24 May 2023].

Bonaccorso, G. (2017b). Machine Learning Algorithms. [online] Google Books. Packt Publishing Ltd.
Available at: https://books.google.com/books?hl=en&lr=&id=_-
ZDDwAAQBAJ&oi=fnd&pg=PP1&dq=machine+learning+algorithms&ots=eplDw_GC5J&sig=PaON0keN
NQrYwpr_iWUTGJbXFp4 [Accessed 24 May 2023].

Carleo, G., Cirac, I., Cranmer, K., Daudet, L., Schuld, M., Tishby, N., Vogt-Maranto, L. and Zdeborová, L.
(2019a). Machine learning and the physical sciences. Reviews of Modern Physics, 91(4).
doi:https://doi.org/10.1103/revmodphys.91.045002.

Carleo, G., Cirac, I., Cranmer, K., Daudet, L., Schuld, M., Tishby, N., Vogt-Maranto, L. and Zdeborová, L.
(2019b). Machine learning and the physical sciences. Reviews of Modern Physics, 91(4).
doi:https://doi.org/10.1103/revmodphys.91.045002.

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23
Jordan, M.I. and Mitchell, T.M. (2015a). Machine learning: Trends, perspectives, and prospects.
Science, 349(6245), pp.255–260. doi:https://doi.org/10.1126/science.aaa8415.

Jordan, M.I. and Mitchell, T.M. (2015b). Machine learning: Trends, perspectives, and prospects.
Science, 349(6245), pp.255–260. doi:https://doi.org/10.1126/science.aaa8415

Pearson BTEC HN RQF Assignment Feedback Form Template Academic Year 2022/23