MNIST Random Forest Classifier

Compulsory assignment for the 2026 Machine Learning course.

Project Overview

This project trains, fine-tunes, and evaluates a Random Forest classifier on the MNIST dataset of handwritten digits. The goal is to classify grayscale digit images from 0 to 9 using a classical machine learning approach.

The project was implemented in Jupyter Notebook using Miniconda and Python libraries from the scikit-learn ecosystem, with Optuna used for hyperparameter optimization.

Notebook

The full work for this assignment is documented in the notebook below:

• mnist_random_forest_classifier.ipynb