Question 6
Domain 2: Data ProcessingWhich feature is the weakest candidate for a direct log transform?
Correct answer: A
Explanation
Direct log transforms require positive values; the guide says to identify scenarios where log scale transformation is appropriate and to “exponentiate log-transformed variables” later, which implies the transform is not suitable for raw zeros or negatives. A feature with “many negative values and zeros with no preprocessing plan” is weakest because log is undefined for nonpositive values unless you first shift or otherwise preprocess the data.
Why each option is right or wrong
A. A feature containing many negative values and zeros with no preprocessing plan
The exam guide’s data-processing objective on log scale transformation is limited to cases where the feature can be validly transformed before later exponentiation of predictions, which presumes nonnegative/positive inputs. Under the standard log definition, log(x) is undefined for x 0 , so a column with many zeros and negative values cannot be directly transformed without an explicit shift or other preprocessing step.
B. A positive monetary amount with a long right tail
Positive monetary amounts with right skew are classic candidates for log scaling.
C. A positive count variable with extreme skew
Positive count variables with extreme skew often benefit from log transformation.
D. A positive variable spanning several orders of magnitude
Positive variables spanning several orders of magnitude are well-suited to log scaling.