One more Question.
Why is accuracy is only 78% ? What should i do for getting more accuracy ? And also , how can i know that what is going behind this algorithm ?
You have taken the mean and std only for trainData(Not all data). But during the normalization you have normalized the 'trainData' as well as 'testData' with same means and std. Why ?
I think you had to calculate the mean and std value for whole data sets.