Next level random forest from scratch

This repo includes code for my Next level random forest from scratch.

Requirements

• Python 3.x
• Numpy
• matplotlib
• joblib

Features and capablities

A classifier built with this library has the following features and capablities:

• Simple, sk-learn like usage form
• Parallel CPU processing through joblib
• Train decision tree or random forest classifiers.
• Ability to limit maximum tree depth for random forest model.
• Compute out-of-bag error.
• Save and load model

Training sample

A full model training code can be found in final_run.py. Dataset used is fashion mnist and can be found in .\input folder.
The final model achieves 86.05% test accuracy.

Sample code

from Classifiers import RandomForestClassifier
import numpy as np
import Preprocessing as pre
import h5py

training_set, test_set = pre.load_data()
X_train, Y_train, X_test, Y_test = pre.preprocess_all(training_set, test_set, -1)
clf = RandomForestClassifier(n_trees=275, tree_depth=50, split_metric='entropy', n_jobs=1)
for i, (x_t, y_t) in enumerate(zip(X_train, Y_train)):
  clf.fit(x_t, y_t)