pytrain

Machine Learning library for python

This library implemented only with python and numpy

Algorithms

• Decision Tree(ID3)
• Gaussian NaiveBayes
• NaiveBayes
• KNN
• Neural Network(FNN)
• Logistic Regression
• Linear Regression
• DBSCAN
• Apriori
• Kmeans
• HierarchicalClustering
• SVM
• SVC (SVM classifier)
• HMM
• CRF

Requirements

• Numpy
• Python 2 or 3

Installation

$ sudo pip install --upgrade pytrain

Basic Usage

import numpy as np
from pytrain.NeuralNetwork import FNN

# Simple dataset
train_mat = [[0.12,0.25],[3.24,4.33],[0.14,0.45],[7.30,4.23]]
train_label = [[0,1],[1,0],[0,1],[1,0]]

test_a = [0.10,0.33]
test_b = [4.0,4.5]

# Train model (FNN)
hidden_layer = [3,2]
fnn = FNN(train_mat, train_label, hidden_layer)
fnn.fit(lr = 0.01, epoch = 2000, err_th = 0.001, batch_size = 4)

# Test model (FNN)
res_a = np.rint(fnn.predict(test_a))
res_b = np.rint(fnn.predict(test_b))

print("X %s => Y %s" % (test_a, res_a))
print("X %s => Y %s" % (test_b, res_b))

———————— output ————————

X [0.1, 0.33] => Y [ 0.  1.]
X [4.0, 4.5] => Y [ 1.  0.]

See more examples here

How to contribute

Fork this repository, and write your algorithm, pull request.
Don't forgot proper test code in test_pytrain.
Test code should be work successfully in below command.

$ python test.py

Reference

• Machine Learning in Action by Peter Harrington (2013)
• Pattern Recognition by Ohilseok (2008)
• Machine Learning to Deep Learning by Deepcumen (2015)
• Pattern Recognition and Machine Learning by Christopher M. Bishop (2006)
• Sequential Minimal Optimization for SVM by John C.Platt (1998)