Shuffle train and test data

[erfannoury]

shuffle data before splitting to test and train subset and train classifiers on the shuffled data.
It will affect classifiers like kNN.