MLA-C01 Data Preparation for Machine Learning Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Replace missing with constant

  Why it's wrong here

  Replacing missing entries with a constant inserts one fixed value into every affected column, so each feature's central tendency is not preserved. It is tempting because constant replacement is the standard choice when a sentinel such as zero or "Unknown" carries genuine business meaning.

• ✗

  Custom transform with Python

  Why it's wrong here

  A custom Python transform would work, but it requires hand-written code to compute per-column medians and assign them, whereas Data Wrangler ships a built-in median imputation transform. Custom transforms suit bespoke logic such as domain-specific conditional fills that no prebuilt handler covers.

• ✗

  Drop missing rows

  Why it's wrong here

  Dropping rows removes the incomplete records entirely, so no median value is ever calculated or substituted, leaving the feature distributions truncated. It is tempting because row deletion is the standard remedy when missingness is rare and random, or when training data must contain only complete cases.

• ✓

  Handle missing values (with median strategy)

  Why this is correct

  The Handle missing values transformation with the median strategy computes each numeric feature's median and substitutes it for nulls, satisfying the requirement to impute with the median rather than mean or mode. It operates directly on the dataset within Data Wrangler's flow.

Go deeper

Related to this question