machine-py 0.0.1

A package for machine learning

MachinePy

A python library based on NumPy, SymPy, Matplotlib.

Introduction

The package has been written in for supervised, unsupervised and semi-supervised machine learning algorithms. All the major basic and advanced machine learning algorithms are written for clustering, rergression, classificaton, dimensionalty reduction problems. Aim of the project is purely educational and it was considered to be optimal while writing.

Functionality

Currently, the release contain several models for the above stated functions. The structure of the proect is as follows:

• Clustering: K-Means,(K-means++), K-Median, (K-Median++), Heirarchical, Mean-Shift, Fuzzy C-Mean,Gaussian Mixture Models and Spectral Clustering.

• Regression: Linear Regreession, Linear Regresson L2 regularised, Linear Regression L1 Regularised, Linear regression L1,L2 Regularised, MLE Linear Regression, Bayesian Ridge Regression, Gaussian Processes, RANSAC, Nadaraya Watson Regression, Local Regression, KNN Regression, Perceptron/ADALINE Regression, Chebyshev-FLNN, Legendre-FLNN, Laguerre-FLNN and Radial Basis Function Neural Net.

• Classification: K-Nearest Neighbors, Logistic Regression, Logistic Regression L2 Regularised, Logistic Regression L1 Regularised, Logistic Regression L1 L2 Regularised, Linear Discriminant Analysis, Quadratic Discriminant Analysis, Naive Bayes, SVM, Perceptron and RBF Neurl Network.

• Dimensionality Reduction: Principal Component Analysis (PCA), Probabilistic PCA, Random Projection, classical Multi-Dimension Scaling(cMDS), LDA, Kernel PCA, Kernel LDA, Isomap and Discriminant Neighborhood Embedding (DNE).

• Semi-Supervised Clustering: Constrained K-means, Seed K-means and COP K-means.

• Semi-Supervised Regression:Co-Training Regression.

• Semi-Supervised Classification:Pseudo Labelling, Cluster & Label, Self Training and Co-Training.

• Semi-Supervised Dimensionality Reduction:SSDR-M, SSDR-CM & SSDR-CMU and SSDR-Manifold.

Installation

Soon to be available thorugh PIP install.

Contributor

Prashant Lawhatre and Pranay Lawhatre

License

BSD 3-Clause License