Contents
DAY 1 – Fundamentals and First Operational Models
Module 1: From Classical Techniques to ML
Limits of traditional expert systems
The three Machine Learning paradigms: Supervised, Unsupervised, Reinforcement
Practical application: Markov Chain for text generation
From determinism to probabilistic predictions
Module 2: Linear Regression
Numerical value prediction: from y = ax + b to the multidimensional world
Standard ML workflow: Problem Definition → Deployment
Practical Project: House Price Predictor with California Housing Dataset
Evaluation metrics: R², RMSE, and business interpretation
Module 3: Classification with Random Forest
Categorical decisions and classification problems
Ensemble Learning: the wisdom of the crowd applied to algorithms
Practical Project: Wine Quality Classifier with UCI dataset
Advanced metrics: Precision, Recall, and business trade-offs
Module 4: Cross-Validation
Avoiding self-deception: overfitting and generalization
K-Fold Cross-Validation for robust estimates
Systematic comparison of different algorithms
Evidence-based decision making
DAY 2 – Advanced Techniques and Production
Module 5: Data Preprocessing
Handling real-world data: missing values, outliers, different scales
Imputation strategies and feature scaling
Automated pipelines with scikit-learn
Preventing data leakage
Module 6: Unsupervised Learning
Discovering hidden patterns: clustering and segmentation
K-Means algorithm and optimization
Practical Project: Customer Segmentation with RFM analysis
Evaluation without ground truth: Silhouette Score and Elbow Method
Module 7: Hyperparameter Tuning
Systematic performance optimization
Grid Search vs Random Search vs Bayesian Optimization
Learning Curves for advanced diagnostics
Balancing overfitting/underfitting
Module 8: Deployment and Production
From Jupyter notebook to production system
Model persistence and versioning
Practical Project: REST API for model serving
Monitoring, A/B testing, and model drift detection
Included Practical Projects
House Price Predictor – Linear regression on California Housing dataset (20k+ samples)
Wine Quality Classifier – Random Forest on UCI dataset with feature importance
Customer Segmentation – K-Means clustering for personalized marketing
Production API – Flask microservice production-ready with error handling
