Overview

This project implements a digit classification pipeline using the MNIST dataset.

Dataset

MNIST consists of 28×28 grayscale images of handwritten digits (0–9).

Approach

• Data loading and visualization
• Feature and target separation
• Data preprocessing (scaling)
• Logistic Regression model training
• Model evaluation

Model & Evaluation

• Model: Logistic Regression
• Metric: Accuracy
• This project uses a simple baseline classifier to demonstrate the full ML pipeline.

Notes

Convolutional Neural Networks typically achieve higher accuracy on MNIST, but this project focuses on interpretability and fundamentals.