A student is using k-nearest neighbour classification to recognize two different image classes: Apples and Oranges. The student has collected 16 images of apples and 17 images of oranges as the training set. The student has found the classification accuracy is very high when k= 3. However, due to a typo in the code, the student actually sets k= 33. Assume that the test set contains 5 images of apples and 5 images of oranges. Comment on the actual classification performance during testing.