Python-package Introduction

This document gives a basic walk-through of BBF Python-package.

List of other helpful links

Install

The preferred way to install BBF is via pip from Pypi:

pip install BBF

To verify your installation, try to import BBF in Python:

import BBF

The BBF Python module can load data from:

NumPy 2D array(s)

import numpy as np
data = np.random.rand(500, 10)
label = np.random.randint(2, size=500)

BBF can use a dictionary to set parameters. For instance:

param = {'max_iterations': 200, 'active_function': 'relu', 'n_nodes_H': 100, 'reg_alpha': 0.001, 'random_state': 0}

Training a model requires a parameter dictionary and data set:

estimator = BBF.BFClassifier(**param).fit(data, label)

After training, the model can be saved:

estimator.save_model('model.joblib')

A saved model can be loaded:

import joblib
estimator = joblib.load('model.joblib')

A model that has been trained or loaded can perform predictions on datasets:

# 7 entities, each contains 10 features
data = np.random.rand(7, 10)
ypred = estimator.predict(data)