Gaussian Naive Bayes Classifier

With Continuous and Discrete or Categorical Functionality

A lean and efficient program for fitting Naive Bayes classifiers to either qualitative or quantitative data. In the latter case, a Gaussian distribution is used to compute probability densities. Some minor refactoring is planned, including addition of other distributions. The functionality for qualitative data has not yet been extensively tested.

Features

• Works with qualitative or quantitative data
• Uses matrix operations wherever possible for efficient operation
• Return predictions as either predicted classes (predict) or posterior probability densities (p_predict).
• Uses log probabilities for quantitative data.
• A scoring function that returns
  • Accuracy
  • Precision
  • Recall
  • F1
  • Easily modified to compute other metrics like specificity or fall-out.

Get the code here