Syllabus

Introduction to Python Programming
Installation of Python and IDEs

Practice control flow, loops, conditionals, functions

Python for Machine Learning
NumPy: Arrays and matrix operations

Matplotlib: Line plots, scatter plots

Pandas: Data loading, summary statistics

Linear Regression from Scratch
Generate synthetic dataset

Implement gradient descent

Compare with Scikit-learn's LinearRegression

Logistic Regression for Classification
Train on binary classification dataset

Perform 5-fold cross-validation

Evaluate with Accuracy, Precision, Recall, F1-score

Analyze results with confusion matrix

Multi-Layer Perceptron (MLP) for Classification
Datasets: Iris and MNIST

Preprocessing: normalization, encoding, splitting

Design MLP using ReLU and Softmax

Train and evaluate model

Visualize training process and results

MLP for Regression Tasks
Work with continuous-valued dataset

Design and train MLP

Use MSE, RMSE, MAE for evaluation

k-NN Classification
Load and explore Iris dataset

Implement k-NN from scratch

Evaluate using accuracy

Clustering with k-Means and k-Means++
Generate data using make_blobs

Apply k-Means and k-Means++

Evaluate using silhouette score

Load Forecasting Using Machine Learning
Use time-series data (real/synthetic)

Handle missing data, normalize

Apply regression models for forecasting

Evaluate with RMSE and MAE

Compare with Scikit-learn models

To equip students with practical skills in implementing machine learning algorithms using Python and popular ML libraries, enabling them to analyze, model, and evaluate data-driven solutions for regression, classification, and clustering problems, including applications such as load forecasting.

By the end of this laboratory course, students will be able to:

Develop proficiency in Python and its libraries essential for data analysis and machine learning.

Implement machine learning models from scratch and using established libraries.

Analyze supervised learning techniques such as regression and classification.

Explore unsupervised learning methods including clustering.

Apply machine learning to real-world problems such as load forecasting.

Evaluate model performance using appropriate metrics and visualization techniques.

Aurélien Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, O'Reilly

Tom M. Mitchell, Machine Learning, McGraw Hill