This function randomly partitions data into training, validation and testing data using Cross Validation. Partitioning data in this manner is commonly used for determining the performance of algorithms with free parameters. Training data is commonly used to train the system, the optimum value for the free parameters is determined using validation data. Finally the results of the algorithm are determined using testing data. Matlab has a cross validation but I find it hard to use so I wrote my own. An example determining the optimum kernel with Support vector machines is used. Contact me if you have questions or suggestions