Question

Consider that you are training a classifier for a 10-class classification problem. Each input is represented as a 512-dimensional vector. There are 1000 samples, out of which the first 100 will be used for testing.
Let Leave-One-Out-Cross-Validation (LOOCV) be used for selection of the classifier model before testing.
Which of the following options is the correct number of validation splits that will be generated?

• A. 10
• B. 512
• C. 900
• D. 1000

Hint:

LOOCV always generates one split per sample in its working pool; here that pool excludes the 100 test samples, leaving 900.

Answer 1

The Correct Option is C

Step 1: Understanding the Question.
We are told 100 out of 1000 samples are set aside purely for final testing, and model selection happens before that testing stage, using LOOCV on whatever samples remain. We need the number of validation splits LOOCV creates.

Step 2: Key Formula or Approach.
Leave-One-Out-Cross-Validation is a special case of k-fold cross validation where \(k\) equals the number of samples in the pool being used. In each split, exactly one sample is held out as the validation point, the model trains on all the rest, and this repeats once for every sample in that pool. So the number of splits generated by LOOCV always equals the size of the dataset it is run on, not the total dataset size, only the portion actually available to it.

Step 3: Detailed Explanation.
The question specifies model selection (which uses LOOCV) happens "before testing", meaning the 100 samples reserved for testing are excluded entirely from the LOOCV process, they are never touched during model selection. That leaves \(1000 - 100 = 900\) samples as the pool LOOCV works on. Since LOOCV produces one split per sample in its working pool, it produces exactly 900 validation splits, one for each of these 900 samples taking a turn as the held-out validation point.

Step 4: Why the other options fail.
Option (A), 10, is the number of classes in the problem, which has nothing to do with how many times LOOCV partitions the data. Option (B), 512, is the input feature dimensionality, also unrelated to the sample count LOOCV cycles through. Option (D), 1000, would be correct only if LOOCV ran over the entire dataset including the 100 test samples, but the question explicitly reserves those for testing only, so they are excluded from the LOOCV pool.

Step 5: Final Answer.
The correct number of validation splits is 900, option (C). \[ \boxed{900} \]