Fashion MNIST PCA Analysis

What this notebook does

This notebook demonstrates dimensionality reduction using Principal Component Analysis (PCA) on the Fashion-MNIST dataset. You'll work through a real-world scenario where you need to reduce 784 pixel features down to a manageable number of components while maintaining model accuracy. The notebook covers:

• Loading and visualizing Fashion-MNIST data (70,000 grayscale images of clothing)
• Training a baseline Random Forest classifier on the full 784-dimensional dataset
• Applying PCA to reduce dimensionality
• Comparing model performance and training time with different numbers of components
• Visualizing data in 2D and 3D using principal components

File structure

pixel-features.ipynb    # Main notebook
data/
  fashion-mnist_train.csv
  fashion-mnist_test.csv

Setup

1. Download the datasets from the Fashion MNIST dataset on Kaggle
2. Place both CSV files in the data/ directory

How to run the notebook

Open pixel-features.ipynb in your editor (VS Code, Jupyter Notebook, or JupyterLab) and run the cells in order.