Very crude implementations of common machine learning algorithms.

These are some basic implementations of machine learning algorithms from scratch, mostly using numpy/scipy. They are designed as a learning experiment and are not suitable for general use other than that.

Currently, this repository contains the following directories:

• supervised/
  • knn.py - K-Nearest Neighbours Classifier
  • simple_linear_regression.py - Linear Regression designed for single dependent and indepent variable
  • gaussian_naive_bayes.py - Gaussian Naive Bayes classifier
  • perceptron.py - Single Layer Perceptron
  • logistic_regression.py - Logistic Regression classifier
  • decision_tree_id3.py - Decision Tree using ID3 algorithm
• unsupervised/
  • kmeans.py - simple KMeans
  • dbscan.py - Density-Based Spatial Clustering of Applications with Noise
  • mean_shift.py - Mean Shift
  • pca.py - Principal Component Analysis
  • rake.py - Rapid Automatic Keyword Extraction
  • spectral_clustering.py - Spectral Clustering
  • yake.py - Yet Another Keyword Extractor
• utils/
  • metrics.py - system evaluation metrics
  • calculations.py - distances and other math related calculations