As I understand, it’s essentially the same as the ‘Akaike infomation criterion’. Searching the File Exchange for ‘AIC’ turns up a number of contributions, the most relevant of which to what you want to do is likely Fast & Detailed Multivariate OLS Regression.
The Mallow’s Cp statistic seems easy enough to program. The complications would show themselves in applying it to your regression — what regressors are kept in and what regressors are omitted from the final model. If you have the Statistics Toolbox, the Stepwise Regression routines do all of that for you.
