http://code.activestate.com/recipes/tags/machine_learning-data_mining/2014-07-31T15:55:15-07:00ActiveState Code Recipes2014-07-31T15:55:15-07:00Chaobin Tang (唐超斌)http://code.activestate.com/recipes/users/4174076/http://code.activestate.com/recipes/578914-simple-linear-regression-with-pure-python/

<p style="color: grey"> Python recipe 578914 by <a href="/recipes/users/4174076/">Chaobin Tang (唐超斌)</a> (<a href="/recipes/tags/algorithms/">algorithms</a>, <a href="/recipes/tags/machine_learning/">machine_learning</a>, <a href="/recipes/tags/python/">python</a>, <a href="/recipes/tags/regression/">regression</a>). </p> <p>Linear regression is a very useful and simple to understand way for predicting values, given a set of training data. The outcome of the regression is a best fitting line function, which, by definition, is the line that minimizes the sum of the squared errors (When plotted on a 2 dimensional coordination system, the errors are the distance between the actual Y' and predicted Y' on the line.) In machine learning, this line equation Y' = b*x + A is solved using Gradient Descent to gradually approach to it. Also, there is a statistical approach that directly solves this line equation without using an iterative algorithm.</p> <p>This recipe is a pure Python implementation of this statistical algorithm. It has no dependencies.</p> <p>If you have pandas and numpy, you can test its result by uncommenting the assert lines.</p>