How to CSS

Build a Random Forest, Today!

Hello! This is a simple python command line thing to build/update the random forest model.

2 Commands

• update-d: updates the default random forest model with the prefined settings
• new: creates a new random forest model

2.1 update-d

2.2 new

• runs build_forest_proto(model_path)
  • model_path: .joblib file (e.g. model.joblib)
• Prompts:
• Partition Size (float): 0.25 - 0.75
• Random Seed (partition) (int): 1 - max(int)
• Number of trees: 1 - max(int) <- don't go max int because it'll be super slow and create a giant file (100,000 trees ~= 2.5gb)
• Criteria (gini or entropy):
  • gini: info gain good for continuous data
  • entropy: info gain good for attributes which occur in classes
• Random Seed (classifier): 1 - max(int)

3 Default Random Forest Classifier Settings:

• Partition Settings:
  • Split Size: 0.5 (50%)
• Classifier Settings:
  • n_estimators: 100
  • oob_score: True