Algorithms and Topics Covered
The course will cover:
Data Processing & Fundamentals
● Data cleaning, preprocessing, and feature engineering
● Train/test splitting, cross-validation, and model selection
● Metrics for regression, classification, and clustering
Classical Machine Learning Algorithms
● Linear Models: Linear regression, ridge/lasso regression, logistic regression
● Nearest Neighbors: k-nearest neighbors for classification and regression
● Decision Trees and Ensembles: CART, random forests, gradient boosting
● Clustering: k-means, hierarchical clustering, DBSCAN
● Dimensionality Reduction: PCA, t-SNE, autoencoders (introductory)
● Naïve Bayes and Probabilistic Models
Neural Networks
● Basics of feed-forward neural networks
● Activation functions, backpropagation, optimization (SGD, Adam)
● Introductory convolutional neural networks (CNNs)
● Regularization (dropout, weight decay)
Additional Methods & Concepts
● Support Vector Machines (SVMs)
● Ensemble learning and stacking
● Introduction to reinforcement learning and generative models (time permitting)
Final Project: Jupyter Notebooks and Machine Learning Package
Objective:
By the end of the semester, students will develop a cohesive body of work that combines Jupyter notebooks demonstrating course concepts with a custom Python machine learning package. This project highlights realworld skills in software development, reproducibility, and communication.
Project Requirements:
● Jupyter Notebooks:
○ Implement a variety of algorithms taught in class
○ Include data exploration, preprocessing, modeling, and evaluation workflows
○ Demonstrate clear explanations using Markdown cells
● Machine Learning Package:
○ Core Functionality: At least two machine learning algorithms implemented as reusable functions or classes
○ Documentation: A README file describing purpose, functionality, and usage
○ Unit Tests: Comprehensive testing with pytest
○ Examples: At least one Jupyter notebook demonstrating the package on a dataset
Submission:
The notebooks and package must be hosted on a public GitHub repository with an organized structure, clear commit history, and detailed documentation.
Grading Criteria
This course is centered around creating a public GitHub repository showcasing knowledge of machine learning and best practices in data science and software engineering. Each student (or small group) will present their repository at the end of the semester, which will account for 100% of the total grade.
● Functionality and Implementation (40%) - Correctness, efficiency, and completeness of algorithms and workflows
● Documentation and Readability (20%) - Quality of README, inline code comments, and Markdown in notebooks
● Testing and Reliability (20%) - Unit tests ensuring code robustness
● Examples and Usability (10%) - Clear, relevant demonstrations of package use
● Repository Quality (10%) - Organization, commit history, and professional practices