This MATLAB program takes as input a collection of input variables called “predictors” and a collection of output variables called “targets” which are arranged in a spreadsheet such that each row of the spreadsheet corresponds to a distinct data record. Two spreadsheets with this format are input to this program. One spreadsheet specifies the “training data”, while one spreadsheet specifies the “test data”. The best minimum mean-square error solution is then computed using a pseudo-inverse least squares solution. Prediction errors for both the “training data” and “test data” are reported based upon estimating the parameters of the linear regression model using the training data. Specific details of this implementation can be found by visiting Podcast Episode 13.
IMPORTANT! This is MATLAB source code. It is necessary to have MATLAB (www.mathworks.com) installed on your machine to run this source code. A public-domain version of MATLAB is available (see: http://www.gnu.org/software/octave) but this MATLAB source code may need to be modified to run in the OCTAVE environment.
