decision_boundary_viz

decision_boundary_viz

A web app for visualizing decision boundaries of classifiers.

Supports 2-dimensional data and the following classifiers and their parameters:

• k-Nearest Neighbors
  • Number of neighbors
  • Weight function
    • Uniform weights
    • Points weighted by the inverse of their distance
• Random forest
  • The number of trees in the forest
  • The maximum depth of trees
• Support vector machine (RBF kernel)
  • Cost (C)
  • Gamma

Demo

Dependencies

• NumPy (numerical operations)
• scikit-learn (classifiers)
• Bokeh (interactive visualizations)
• Flask (web framework)

conda create -n decision_boundary_viz python=3.5 numpy=1.11 scikit-learn=0.18 bokeh=0.12.4 flask=0.12