MSc Project Proposal Form

Student number 2183337

Student name Sandeep Singh

Course name Msc Computer Science

Supervisor name Piotr Wojtasiuk

Project title Big Data Analytics in Retail: Customer Segmentation, Demand

Forecasting, and Inventory Management

Section 1: Project Proposal

Description of your artefact

Description of Artefact

Within the context of the project, a detailed Big Data analytics portfolio for retail will be created; its
key components will include customers’ clustering, demand forecasting, and inventory modelling. It
can utilize the measures of big data analytics and various algorithms by incorporating in its body
artificial intelligence to make informed decisions and enable efficiency within the retail industry.

Context of Project

Some issues mentioned above are addressed by existing solutions in retail analytics in a way, or not
addressed at all integrated. Even today’s offerings from python for retail are strong support for
analytics, but there could be a lack of compatibility or live monitoring across all three targeted
regions. Thus, this project seeks to solve by developing a solution that offers the solution for every
domain in detail while integrating them into one harmonized module.

Aim & Objectives

Aim: For instance, using Big data to create an integrated big data analytics platform for customer
segment, demand forecast and inventory management in sectors such as Retail.

Objectives:

Deploy application modules based on clustering of customers.

Ensure creation of prognostic models that would help in demand forecasting.

Intelligent management of inventory to minimize the amount of stock by introducing a more effective
inventory control system.

Organize these modules in one place that will be easy to use for the majority of the users.

Check the platform’s applicability also through lab experiments and practical examples.

Features of the Artefact

Customer Segmentation: Involve clustering methods (K-means, Hierarchical etc) to segment customer
in terms of purchasing ability and other related factors.

Demand Forecasting: Uses statistical and prognostic techniques such as time series forecasting and
machine learning methods such as ARIMA models and LSTM for future product sales forecasts.

Inventory Management: Has incorporated optimization algorithm to manage its inventory while at the
same time trying to avoid incidences of stockouts while containing costs of holding inventory.

Real-Time Analytics: It is particularly giving the data processing and data visualization to the real
time data.

User Interface: Alternative, users should be able to have an embedded and easy to use dashboard to
interact with the data and insights to support their decision-making.

Added Value

They combined three important functions in retailing in one place in the project which improves
efficiency in making decisions. In contrast to the existing tools, it is simple yet effective with focus on
the real-time analysis and integration that can benefit retailers.

Intellectual Challenges

Statistical modelling used in demand forecasting so as to come up with accurate and more robust
models.

Real-time data processing and the integration of facilities among various modules must be ensured.

Designing a conveniently convenient package that can read and precondition large and complicated
data sets.

What methodology (structured process) will you be following to realise your artefact?

Development Approach

The project will be developed under the Agile notion with significant focus on the iterative and
increment approach and constant feedback. This approach is most suitable for big data analytics
projects because the changes can be made had and often depending on the feedback from the
application.

Initiation and Planning

Stakeholder Engagement: Start from detailed meetings with a view of having comprehensive
understanding of the potential clients’ needs before finalizing their needs to conform the business
objectives.

Literature Review and Technology Selection: Research doing I R & D on all currently available and
practicable methods and technologies in retail analytics particularly on customer segmentation,
demand forecast and inventory management. Choose proper tools and technologies for this research
setting based on the research.

Design
System Architecture: Develop a modular system-driven system architecture that allows for relatively
easy integration of the three fundamental business modules that deal with customer segmentation
data, forecasting of demand, and inventory management.

Module Specifications: Determining the algorithms for the modules as well as what data would be
required to support the whole module.

Development

Customer Segmentation Module: Regarding the development of this module, clustering algorithms
such as K-means and hierarchical clustering should be employed for estimating the assigned value
based on customer behaviour and demographic information. This comprises data cleaning, feature
extraction, algorithm specification, and model training.

Demand Forecasting Module: After, use Historical Time Series Analysis with the ARIMA technique
and Machine Learning with the Long Short Term Memory (LSTM). Some of these will be feature
selection and creation, model training, and efficiency assessment over previous data sets.

Inventory Management Module: Design an optimization based system for performing inventory
control in order to maintain satisfactory stock conditions. This module will be particularly designed in
a manner that it does not allow the company to incur in high holding costs of inventory as well as
experiencing stock out situations by forecasting the demand of inventory.

Integration

Unified Platform Development: Conduct the scientific evaluation of the three modules as a set and
implement them in one system. Ensure that there is an interface that can enable users to engage with
the collected information without distortions.

Real-Time Processing: Achieve real-time data processing so that information is accurate and timely to
meet the needs of the organization.

Testing and Evaluation

System Testing: Perform an effective unit testing of each individual module and of integrated system
altogether. Employ real-life statistics to guarantee that all information in the solution is clear and
trustworthy.

User Feedback: It is recommended users participate during tests to be used in assessment of the tests,
with particular focus on usability and functionality. If possible, adjust according to the feedback that
was given.

Performance Evaluation: Analyse the results to discover how well the platform works according to
accuracy, precision, recall as well as customers’ satisfaction.

Research Methodology

Data Collection: Collect data either from retailers or other available public data sources to train and
evaluate the models on. Preprocessing the data before its organizational also have steps which
enhance the quality of data.

Dataset
Dataset: The following dataset from Kaggle will be used: Superstore Sales Data set with large sales
data for the analysis. It includes several variables such as OrderID, ProductID, CustomerID, Sales,
Profit, Category, and OrderDate which play the role of dimensions and thus offer significant scope for
analysis and evaluation.

Source: Kaggle Superstore Sales Data Set

(https://www.kaggle.com/datasets/ishanshrivastava28/superstore-sales?select=Superstore.csv)

Model Development: The statistical methods and machine learning algorithms give more reliability
when using them to derive models. There is need to modify the developed models in phases
depending on the performances and reviews indicated.

Evaluation Metrics: Evaluate the suitability of the models and the assessment platform based on the
results yielded by the following metrics. These include accuracy for demand forecasting, cluster
cohesion for customer classification, and inventory turnover ratios for inventory control.

Project Management Approach

In this project, the Agile culture of development will be adopted due to its ability to create and
manage sprints of work. Companies will also combine daily scrum meetings with proper sprint
reviews and retrospectives to guarantee responsiveness and adaptability to changes. It is possible to
rely on this approach to develop the product rapidly, to return to the process frequently, and to deliver
it in a short time while being of a high quality.

How does your project relate to your degree course and build upon the units/knowledge you
have studied/acquired

The project relates to the knowledge and skills attained in Data Science, Machine Learning, Database
Management System, and Computer Software Engineering. It encompasses principles like data
mining, predictive analysis, pattern algorithms, integration systems, among others.

What are the main contributions of your project as compared to state-of-the-art?

Main Contributions

Integrated Platform: Comes with thoughts of customer classification, demand prediction and
inventory positioning as well.

Real-Time Analytics: Performs the computations in real-time, aiding in accelerated decision making
processes and improved decision-making.

User-Friendly Interface: Make it simple, so that the retailer will be able to take advantage of the
myriad calculations involved without the need for a deep technological background.

Resources

Programming Language: Python

Data Manipulation and Analysis: Pandas, NumPy

Data Visualization: Matplotlib, Seaborn

Machine Learning: Scikit-Learn

Time Series Analysis: stats models

Bibliography

Chen, I.F. and Lu, C.J., 2021. Demand forecasting for multichannel fashion retailers by integrating
clustering and machine learning algorithms. Processes, 9(9), p.1578.

Ghalehkhondabi, I., Ahmadi, E. and Maihami, R., 2020. An overview of big data analytics application
in supply chain management published in 2010-2019. Production, 30, p.e20190140.

Gopal, P.R.C., Rana, N.P., Krishna, T.V. and Ramkumar, M., 2024. Impact of big data analytics on
supply chain performance: an analysis of influencing factors. Annals of Operations Research, 333(2),
pp.769-797.

Kumar, A., Shankar, R. and Aljohani, N.R., 2020. A big data driven framework for demand-driven
forecasting with effects of marketing-mix variables. Industrial marketing management, 90, pp.493-
507.

Punia, S., Nikolopoulos, K., Singh, S.P., Madaan, J.K. and Litsiou, K., 2020. Deep learning with long
short-term memory networks and random forests for demand forecasting in multi-channel
retail. International journal of production research, 58(16), pp.4964-4979.

Raji, M.A., Olodo, H.B., Oke, T.T., Addy, W.A., Ofodile, O.C. and Oyewole, A.T., 2024. Real-time
data analytics in retail: A review of USA and global practices. GSC Advanced Research and
Reviews, 18(3), pp.059-065.

Saha, P., Gudheniya, N., Mitra, R., Das, D., Narayana, S. and Tiwari, M.K., 2022. Demand
forecasting of a multinational retail company using deep learning frameworks. IFAC-
PapersOnLine, 55(10), pp.395-399.

Seyedan, M. and Mafakheri, F., 2020. Predictive big data analytics for supply chain demand
forecasting: methods, applications, and research opportunities. Journal of Big Data, 7(1), p.53.

Tian, X., Wang, H. and Erjiang, E., 2021. Forecasting intermittent demand for inventory management
by retailers: A new approach. Journal of Retailing and Consumer Services, 62, p.102662.

Zineb, E.F., Najat, R.A.F.A.L.I.A. and Jaafar, A.B.O.U.C.H.A.B.A.K.A., 2021. An intelligent

approach for data analysis and decision making in big data: a case study on e-commerce
industry. International Journal of Advanced Computer Science and Applications, 12(7).

Section 2: Project Plan and Gantt Chart

Phases/Sub-phases 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1. Initiation X
2. Research and Planning X
3. Design X X
4. Development X X X X X X X X X
4.1 Customer Segmentation X X X
4.2 Demand Forecasting X X X X
4.3 Inventory Management X
Section 3: Project Risk Assessment

Risk Impact Likelihood Mitigating Action

Secure data sources early, use synthetic
Data Availability High Likely
data for testing.
Iterative testing, refine models
Model Accuracy Moderate Likely
continuously.
Conduct regular integration tests, use
Integration Issues High Likely
modular design.
Involve users in design process, ensure
User Adoption Moderate Likely
user-friendly interface.
Regular technical reviews, consult with
Technical Challenges High Likely
experts.
Efficient resource planning, prioritize
Resource Constraints Moderate Likely
critical tasks.
Security and Privacy Implement strong data encryption and
High Likely
Concerns access controls.