Random-brain

About

Random brain is the neural network implementation of a random forest. Its purpose is to combine the strengths of multiple nerual networks.

Background on random forests

A random forest is a machine learning model that is composed of multiple decision trees. These trees in the forest all predict an outcome and the majority rules.

Similarities

Just as the random forest is a vote based ML algorithm, the random brain is a vote based algorithm as well, but uses neural networks specified by the user rather than decision forests.

Setting up Random brain

pip install random-brain

API

Init the brain module and class.

from random_brain import random_brain
brain = random_brain.random_brain()

import models()

Import models will take in a directory or a single .h5 file. Sub directories will be ignored.

brain.import_models(model_path = 'path/to/model.h5')
brain.import_models(model_path = 'path/to/directory')

show_brain()

Shows the keys used in the brain. This should just be the name of each imported model

brain.show_brain()

clear_brain()

Clear a single model or more by entering in the model name as a list. Leave blank to clear all models.

brain.clear_brain(item_list = ['model to remove'])

vote

Add in yTest to cast votes. Vote() will only return the votes as a numpy array and not actual predictions. This is useful if you want to run your own statistics on the votes.

brain.vote(yTest = [1, 2, 3, 4, ...])

predict (in development)

Add in your yTest to make predictions. This will attempt to make a prediction based off of the networks and will return a single answer. This is still in development.

In the future prediction and threading options will be added and improved.

brain.predict(yTest = [1, 2, 3, 4, ...])