ml-python

The easiest way to do machine learning


Keywords
convolutional-neural-networks, gradient-descent, k-means, learn-machine-learning, library, linear-regression, logistic-regression, machine-learning, ml-python, neural-network, nonlinear-regression, python, tensorflow
License
MIT
Install
pip install ml-python==2.2.2

Documentation

Build Test Downloads PyPi Version License

ML

This module provides for the easiest way to implement Machine Learning algorithms. It also has in-built support for graphing and optimizers based in C.

Learn the module here:

This module uses a tensorflow backend.

Implemented Algorithms

  • 2D CNN ml.cnn
  • Basic MLP ml.nn
  • K-Means ml.k_means
  • Linear Regression ml.linear_regression
    • optimized with C
  • Logistic Regression ml.logistic_regression
  • Graph Modules ml.graph
    • Graph any function with or without data points - from ml.graph import graph_function, graph_function_and_data
  • Nonlinear Regression ml.regression
  • Optimizers - ml.optimizer optimized with C
    • GradientDescentOptimizer - from ml.optimizer import GradientDescentOptimizer
    • AdamOptimizer - from ml.optimizer import AdamOptimizer
  • UNSTABLE - Character generating RNN - ml.rnn

You can find examples for all of these in /examples

Pip installation

pip install ml-python

Python installation

git clone https://github.com/vivek3141/ml
cd ml
python setup.py install

Bash Installation

git clone https://github.com/vivek3141/ml
cd ml
sudo make install

Examples

Examples for all implemented structures can be found in /examples.
In this example, linear regression is used.

First, import the required modules.

import numpy as np
from ml.linear_regression import LinearRegression

Then make the required object

l = LinearRegression()

This code below randomly generates 50 data points from 0 to 10 for us to run linear regression on.

# Randomly generating the data and converting the list to int
x = np.array(list(map(int, 10*np.random.random(50))))
y = np.array(list(map(int, 10*np.random.random(50))))

Lastly, train it. Set graph=True to visualize the dataset and the model.

l.fit(data=x, labels=y, graph=True)

Linear Regression

The full code can be found in /examples/linear_regression.py

Makefile

A Makefile is included for easy installation.
To install using make run

sudo make

Note: Superuser privileges are only required if python is installed at /usr/local/lib

License

All code is available under the MIT License

Contributing

Pull requests are always welcome, so feel free to create one. Please follow the pull request template, so your intention and additions are clear.

Contact

Feel free to contact me by: