scikit-learn — Practical Machine Learning Examples

This repository is a hands-on collection of Python and scikit-learn notebooks. It covers core estimators, preprocessing, model selection, evaluation, model persistence and a complete Bengaluru house-price prediction project.

The original notebooks, datasets, model files and client/server project files are retained. Some examples originate from older tutorial workflows; read the explanations added around them and pay attention to any compatibility notes before running older cells with a newer scikit-learn release.

Learning path

Stage	Existing folders	Key skills
Ensembles and trees	01–02, 14	Bagging, decision trees, random forests
Categorical data	03	Encoding and dummy variables
Model selection	04, 06, 17	Hyperparameter search, cross-validation, train/test split
Clustering and neighbors	05, 07	K-Means, KNN
Regression and classification	08–11	Linear/logistic regression, multivariate regression, Naive Bayes
Dimensionality and generalization	12, 15–16	PCA, regularization, SVM
Applied project	13	Bengaluru house-price workflow and prediction service

Folder guide

1. Bagging
2. Decision Tree
3. One Hot Encoding
4. Hyperparameter Tuning
5. K-Means Clustering
6. K-Fold Cross Validation
7. K-Nearest Neighbor Classification
8. Linear Regression
9. Logistic Regression
10. Multivariate Regression
11. Naive Bayes
12. Principal Component Analysis
13. Projects — Bengaluru house prices
14. Random Forest
15. Regularization
16. Support Vector Machine
17. Train And Test Split
18. Miscellaneous Files

How to study

• Read the explanation before running each code cell.
• Identify what each variable represents and inspect its shape.
• Understand which columns are features and which column is the target.
• Understand whether the task is regression, classification or clustering.
• Check how training and test data are separated.
• Interpret metrics instead of simply trying to maximize a score.
• When a model is saved, understand which preprocessing steps and feature order it expects.

Setup

python -m venv .venv
# Windows PowerShell:
.venv\Scripts\Activate.ps1
# macOS / Linux:
source .venv/bin/activate
python -m pip install --upgrade pip
pip install numpy pandas matplotlib seaborn scikit-learn jupyter joblib
jupyter notebook

Older notebooks may use APIs or prediction formats that generate warnings in recent releases. The correct long-term pattern is to keep feature names and column order consistent between fit and predict, and to use Pipelines for preprocessing plus the estimator.

Important terminology

• Estimator: a scikit-learn object that learns from data through fit.
• Transformer: an object that learns a data transformation and applies it through transform.
• Predictor: an estimator that can produce predictions using predict.
• Pipeline: a sequence of transformers and a final estimator treated as one workflow.
• Hyperparameter: a setting chosen outside the model's learned coefficients, such as max_depth.
• Cross-validation: repeated train/validation splits used to estimate how a model may generalize.
• Feature encoding: converting categories such as city names into numeric representations a model can process.

Project: Bengaluru house prices

The project demonstrates loading and cleaning real-estate data, engineering features such as BHK and price per square foot, handling location categories, training a regression model, tuning alternatives, saving a model and exposing prediction logic through a small server/client example.

The dataset is historical and the resulting model is an educational example. Its predictions must not be treated as current property valuations or financial advice.

Further reading

• scikit-learn User Guide
• scikit-learn API reference
• GeeksforGeeks scikit-learn tutorials
• Pandas documentation
• Matplotlib documentation

Open study-tracker.html to track your study progress.