The "columns" represent the "features" you have extracted (These are the predictors). 

A "response variable" is the class label which you need to include for classification.

So, before you open the app, concatenate the final features with the labels.

For eg: You have the final feature matrix having size = 10 x 3. 

This means that you have 3 features for each of the 10 images/subjects.

(These 3 features will appear in the app as col_1,col_2, col_3).

Now, add another column to this 10 x 3 matrix. This column should contain the class labels for the 10 images/samples. (This will be col_4 in the app, select it as the response variable). This will make your matrix 10 x 4. Open the app next, and select the response variable as col_4).

Abhilash K Pai  Sir Please help me I am confused how 10 images when 10 x 3 ?

feature 1 (colu _1) contains 10 rows then theses will consider as 10 images????????

"feature 1 (colu _1) contains 10 rows "-  This means that the feature which you have extracted (for eg: orientation) has different values for 10 different images (rows).