Skip to content

Latest commit

 

History

History
58 lines (34 loc) · 2.43 KB

README.md

File metadata and controls

58 lines (34 loc) · 2.43 KB
Raw

Non-parametric regression

Change to test travis.

Build Status Codacy Badge

This package provides a set of non-parametric methods for regression. The implementation of the methods are in C++ npr and Python interface is available pynpr. This library is built on FLANN which is a fast approximate nearest neighbour algorithm.

Dependencies

sudo apt-get install libboost-python-dev libflann-dev libarmadillo-dev python-numpy

Installation

  1. Checkout the repository
  clone https://github.com/gpldecha/non-parametric-regression.git
  1. Run the install.py script (from within the folder non-parametric-regression)
  python install.py

The script will check the dependencies and install them if missing and then will first build and install the npr C++ library and then build and install the python interface library pynpr.

Examples

  1. examples/python

A set of python interface examples for 1D, 2D and 3D regressions.

  1. examples/cpp

    A set of examples on how to use the C++ interface.

Regression methods

  • Locally weighted regression (LWR)

Library of locally weighted (LWR) regression. LWR is a memory based regression which is non-parameteric. It stores a set of input, X, and target values, y which form the desired training data for the regressor function y = f(X).