Scribd
Upload
31 views21 pages

Fundamentals of ML Recap

Uploaded by

kayforts
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
31 views21 pages

Fundamentals of ML Recap

Uploaded by

kayforts
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 21

IDEAS Emerging Technology

Skills Scholarship Program


RECAP OF FUNDAMENTALS OF
MACHINE LEARNING
Presented by: Khadijah Saad Mohammed
Content
Recap
Additional Code
We'll review the key concepts and
Recap techniques we've covered in this module,
ensuring you have a solid foundation in
machine learning.
Introduction
Types of Machine Learning
Common Algorithms
Preparing Data and Engineering Features
Evaluating Model Performance
Machine Learning

Machine learning is a subset of artificial


intelligence that involves training algorithms to
predict outcomes, identify patterns, and make
decisions based on data.

CULMINATECH
Types of Machine Learning

Supervised Learning: Where the model learns from labeled data.


Unsupervised Learning: Where the model identifies patterns in
unlabeled data.
Reinforcement Learning: Where an agent learns to make decisions by
performing actions and receiving rewards.

CULMINATECH
Types of Machine Learning
Supervised Learning
Where the model learns from labeled data.
In supervised learning, the model uses input-output pairs to
learn a function that can predict outcomes for new data.
Classification: Predicting categorical labels (e.g., spam
or not spam). Apple Banana

Regression: Predicting continuous values (e.g., price of a


house).
.

CULMINATECH
Types of Machine Learning
Unsupervised Learning

Unsupervised learning involves models that infer patterns


. The model has to make sense of the
from unlabeled data without reference to known or labeled
data on its own, finding patterns and
outcomes. structures that we might not
Clustering: Grouping a set of objects in such a way immediately see.
that objects in the same group are more similar to
each other than to those in other groups.
Association: Discovering rules that describe
portions of your data, such as people that buy X also
tend to buy Y.

CULMINATECH
Types of Machine Learning

Reinforcement Learning
:Where an agent learns to make decisions by performing actions and receiving rewards.

CULMINATECH
Preparing Data and Engineering Features

Data Pre-processing
Cleaning: Removing errors and inconsistencies from the data.
Transformation: Scaling or normalizing features to ensure uniformity.
Encoding: Converting categorical variables into numerical
representations.
Feature Engineering: Creating new features or transforming existing
ones to improve model performance.

CULMINATECH
Model Evaluation
Why Evaluate a Model?
After training a model, you need to know how well it predicts
new data

Basic Metrics In classification, the data is split


For Classifying Data (e.g., spam or not spam): Look at into training sets for learning and
Accuracy (how many predictions were correct). testing sets for evaluating model
For Predicting Values (e.g., house prices): Use MSE (Mean performance
Squared Error), which tells you how far off your predictions
are on average.
Model Evaluation
Evaluating a model's accuracy and performance is crucial for verifying its effectiveness.
Key Metrics
For Classification: Accuracy, Precision, Recall, and F1 Score.
For Regression: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and
Mean Absolute Error (MAE).
Validation Techniques
Train/Test Split
Cross-Validation: Using parts of the data to train and validate the model to ensure
reliability.
Overfitting and Underfitting
What are These?
Overfitting: Imagine memorizing answers for a test without understanding the
concepts. You might fail on different questions. Overfitting is similar; the model
performs well on training data but poorly on unseen data.
Underfitting: This is like not studying enough. Underfitting means the model is
too simple to learn the underlying pattern of the data.
Overfitting and Underfitting
Overfitting: Occurs when a model learns the training data too well, including the
noise, which hampers its performance on new data.
Underfitting: Occurs when a model is too simple to capture the underlying data
patterns, resulting in poor performance on both training and new data.
Solutions:
Regularization: Techniques like L1 (Lasso) and L2 (Ridge) that help reduce
overfitting by penalizing large coefficients.
Pruning: Used in decision trees to reduce the size of the tree and improve model
simplicity.
Regularization Techniques
What is Regularization?

Think of regularization as a way to prevent your model from studying "too hard" and just
memorizing data. It gently nudges the model to be more general.
Best Practices in Machine Learning
Data Quality: High-quality data is critical for building robust models.
Algorithm Selection: There is no one-size-fits-all algorithm. Testing different algorithms based
on the problem context is essential.
MORE CODING
The next slides contain snippets of more coding examples.
from sklearn.linear_model import LogisticRegression
model = LogisticRegression()
model.fit(X_train, y_train)

predictions = model.predict(X_test)
from sklearn.metrics import accuracy_score
print("Accuracy:", accuracy_score(y_test, predictions))
More Data Processing
# Create a new column 'is_child' to indicate whether the passenger is a child
df['is_child'] = (df['Age'] < 18).astype(int)

# Fill missing values


df['Age'].fillna(df['Age'].median(), inplace=True)
df['Embarked'].fillna(df['Embarked'].mode()[0], inplace=True)
# Extract title from the Name column
df['Title'] = df['Name'].str.extract(' ([A-Za-z]+)\.', expand=False)

# Convert categorical variables using one-hot encoding


df = pd.get_dummies(df, columns=['Sex', 'Embarked', 'Title'], drop_first=True)

# Drop columns that are not needed


df.drop(['PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1, inplace=True)
# Creating a new feature 'FamilySize'
titanic['FamilySize'] = titanic['SibSp'] + titanic['Parch'] + 1
Other Algorithms
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC
Accuracy with other
# Logistic Regression
log_reg = LogisticRegression()
Algorithms
log_reg.fit(X_train, y_train)
log_reg_preds = log_reg.predict(X_test)
print("Logistic Regression Accuracy:", accuracy_score(y_test, log_reg_preds))

# Random Forest Classifier


rf_clf = RandomForestClassifier()
rf_clf.fit(X_train, y_train) # no need to scale data for tree-based models
rf_preds = rf_clf.predict(X_test)
print("Random Forest Accuracy:", accuracy_score(y_test, rf_preds))

# Support Vector Machine


svm = SVC()
svm.fit(X_train, y_train)
svm_preds = svm.predict(X_test)
print("SVM Accuracy:", accuracy score(y test, svm preds))
THANK YOU
Q&A

You might also like