While building a predictive model, median imputations are performed while preparing the training data.
How should the imputations be addressed in the validation data?
While building a predictive model, median imputations are performed while preparing the training data.
How should the imputations be addressed in the validation data?
When performing median imputations for missing data during the preparation of training data, it is crucial to apply the same imputed values directly to the validation data without recalculation. This ensures consistency in the model's evaluation, as recalculating the medians using the validation data would introduce information leakage and compromise the validation process.
Correct answer is B. Medians should come from the training data set. This is addressed in SAS' course video Predictive Modeling Using Logistic Regression (15.1) Lesson 4.2 - "...So in the validation data set, missing values should be replaced with the medians from the training data set. " Furthermore, in the quiz at the end of the SAS lesson, one question states a particular answer was incorrect because "Answer b is incorrect because the missing values in the validation data set need to be replaced with the medians from the training data set." Please fix this for others.
Correct answer is B