CODEWITHSUNDEEP
python-3.xnumpyk-fold
Shubham Bajaj
Shubham Bajaj

Reputation: 329

Creating Kfold cross validation set without sklearn

I am trying to split my data into K-folds with train and test set. I am stuck at the end:

I have a data set example:

       [1,2,3,4,5,6,7,8,9,10]

I have successful created the partition for 5-fold cross validation and the output is 

     fold=[[2, 1], [6, 0], [7, 8], [9, 5], [4, 3]]

Now I want to create K such instances having K-1 training data and 1 validation set.

I am using this code:

    ```
      cross_val={"train":[],"test":[]}
       new_fold=folds.copy()
       for i in range(4):
           val=folds.pop(i)
           cross_val["train"].append(folds)
           cross_val["test"].append(val)

           folds[i:i]=[val]```

The output that I am getting is:

  {'train': [[[6, 0], [7, 8], [9, 5], [4, 3]],
           [[6, 0], [7, 8], [9, 5], [4, 3]],
          [[6, 0], [7, 8], [9, 5], [4, 3]],
        [[6, 0], [7, 8], [9, 5], [4, 3]]],
   'test': [[6, 0], [7, 8], [9, 5], [4, 3]]}

This is the wrong output that I am getting.

But I want the output as

          train                                   test
          [[6, 0], [7, 8], [9, 5], [4, 3]]      [2,1]
          [[2, 1], [7, 8], [9, 5], [4, 3]]      [6,0]
          [[6, 0], [2, 1], [9, 5], [4, 3]]      [7,8]
          [[6, 0], [7, 8], [9, 5], [2, 1]]       [4,3]
          [[6, 0], [7, 8], [2, 1], [4, 3]]       [9,5]

Upvotes: 1

Views: 11884

Answers (1)

willeM_ Van Onsem
willeM_ Van Onsem

Reputation: 477794

You here each time make edits to the same list, and append that list multiple times. As a result if you edit the list, you see that edit in all elements of the list.

You can create a cross-fold validation with:

train = []
test = []
cross_val={'train': train, 'test': test}
for i, testi in enumerate(fold):
    train.append(fold[:i] + fold[i+1:])
    test.append(testi)

For the given sample data, this gives us:

>>> pprint(cross_val)
{'test': [[2, 1], [6, 0], [7, 8], [9, 5], [4, 3]],
 'train': [[[6, 0], [7, 8], [9, 5], [4, 3]],
           [[2, 1], [7, 8], [9, 5], [4, 3]],
           [[2, 1], [6, 0], [9, 5], [4, 3]],
           [[2, 1], [6, 0], [7, 8], [4, 3]],
           [[2, 1], [6, 0], [7, 8], [9, 5]]]}

Upvotes: 2

Related Questions