This m-file computes an approximate determination coefficient (R2) for prediction, say
R2pred = 1 - (PRESS/SST)
where PRESS = prediction error sum of squares, SST = total sum of squares.
This statistic gives some indication of the predictive capability of the regression model. So, we could expect the model 'explain' about a percentage of the variability in predicting new observations, as compared to the approximately percentage of the variability in the original data explained by the least squares fit.
Syntax: function x = apR2p(D)
Inputs:
D - matrix data (=[X Y]) (last column must be the Y-dependent variable; X-independent variables).
Output:
x - approximate determination coefficient (R2) for prediction. |
