ISLP Machine Learning Exercises

Ler em Português

This repository contains solved classification exercises using Machine Learning models in Python with scikit-learn and the ISLP textbook package. The source code explores different classification algorithms and the effects on statistical accuracy when separating Train/Test data vs using the full dataset.

---

Code Overview

The main script exercicios_logistica.py consists of 3 practical parts:

1. Random Forest and Breast Cancer Dataset: Loads native sklearn data, trains the model using an 80%/20% split, plots a confusion matrix heatmap, and outputs a complete classification report.
2. Logistic Regression with Train/Test Split: Re-implements 4 classic exercises from the ISLP library, dividing the modeling with independent samples and validating on unseen slices of the table using binarization matrix notation (get_dummies).
   • ISLP::Default (Target: student)
   • ISLP::Smarket (Target: Direction)
   • ISLP::Weekly (Target: Direction)
   • ISLP::Caravan (Target: Purchase)
3. Basic Logistic Regression: Removes train_test_split to train and verify accuracy on the entire dataset, illustrating the statistical concept of "training error underestimation" when predicting samples the model has already seen.

---

Libraries & Installation

To run this project, make sure you have Python 3.10+ and execute the command below in your terminal to fetch the core requirements:

pip install ISLP pandas scikit-learn matplotlib seaborn

Once done, run the main file:

python exercicios_logistica.py

(Note that the Logistic Regression reports from Part 2 and 3 will print in the terminal only after closing the Matplotlib figure window that opens in Part 1).