/discriminant

Trying to learn LDA and QDA

Primary LanguagePythonMIT LicenseMIT

Linear and Quadratic Discriminant Analysis

Implemented these two Machine Learning algorithms while studying CS 189 @ UC Berkeley.

Project Outline:

  • classifier.py - machine learning code of LDA and QDA
  • train.py - sample training code using LDA or QDA with MNIST or SPAM
  • data/spam.mat - matlab file containing spam training data
    • each row represents an email labeled as either "spam" or "ham"
    • each col represents frequency of a word (actual data is l1-normalized)
  • data/spam.obj - python pickle file containing dictionary of words
  • data/mnist.mat - matlab file containing mnist digit training data
    • each row represents a 28-by-28 flatten greyscale image
    • each col represents greyscale from 0 to 255 (actual data is l2-normalized)

The data files are not included in this github repo, they can be downloaded here.