Decision Tree Classification

This repository contains a Jupyter Notebook that implements a Decision Tree classifier on a small social network ads dataset. The notebook follows a standard machine learning workflow: data loading, preprocessing (feature scaling), training, prediction, evaluation (confusion matrix & accuracy), and visualization of the decision boundary for the training and test sets.

Contents

• decision_tree_classification.ipynb — main Jupyter Notebook with the full implementation and visualizations.
• Social_Network_Ads.csv — dataset used by the notebook (features: Age, EstimatedSalary, etc.).

Overview

The notebook demonstrates how to:

• Load a CSV dataset with pandas.
• Split data into training and test sets using train_test_split.
• Apply feature scaling with StandardScaler.
• Train a DecisionTreeClassifier (criterion: entropy).
• Make predictions and evaluate the model with a confusion matrix and accuracy score.
• Visualize decision regions for training and test sets using Matplotlib.

Dependencies

Recommended Python packages (tested with Python 3.8+):

• numpy
• pandas
• matplotlib
• scikit-learn
• jupyter

You can install these packages with pip. Example (PowerShell):

python -m venv .venv; .\.venv\Scripts\Activate.ps1
pip install --upgrade pip
pip install numpy pandas matplotlib scikit-learn jupyter

If you plan to upload this repository to GitHub, consider adding a requirements.txt file generated with:

pip freeze > requirements.txt

How to run

1. Open a PowerShell terminal in this project folder (where the notebook and CSV file are located).
2. Activate your virtual environment (see commands above).
3. Start Jupyter Notebook or JupyterLab:

jupyter notebook
# or
jupyter lab

4. Open decision_tree_classification.ipynb and run the notebook cells in order.

Notes:

• The notebook expects Social_Network_Ads.csv to be in the same folder as the notebook.
• The notebook scales features before training; when calling classifier.predict() for single samples, the same StandardScaler instance is used via sc.transform().

Expected outputs

• A printed confusion matrix and an accuracy score for the test set.
• Two Matplotlib plots showing the decision regions (training and test sets) with the data points overlaid.
• Example single-sample prediction printed to the notebook output.

Quick tips

• If the decision region plots look distorted, check the ranges used to build the meshgrid in the visualization cells. The code uses inverse-transformed values to make the plots show real feature scales.
• If you get a ValueError when predicting on a single sample, confirm you passed a 2D array (e.g., [[30, 87000]]) and that you applied sc.transform() before prediction.

License

This project is provided as-is for educational purposes. Feel free to fork, modify, and use it. If you plan to redistribute, please include attribution.

Contributing

Small improvements are welcome (README tweaks, adding a requirements.txt, or a short script to run the notebook automatically). If you'd like, open an issue or submit a pull request after uploading to GitHub.

---

Happy experimenting with Decision Trees! If you want, I can also:

• add a requirements.txt for you, or
• create a short script that runs the notebook and exports the outputs as images.

Just tell me which one you'd like next.