Question

A data scientist is building a Spark ML pipeline for a churn classification model. They add a StringIndexer, a VectorAssembler, and a LogisticRegression stage. When they call fit() on the pipeline, they want to understand which stages actually learn something from the data versus which simply apply a fixed rule. Which statement correctly describes the behavior of these stages?

Accepted Answer

StringIndexer and LogisticRegression are estimators that learn from data during fit(), while VectorAssembler is a transformer that applies a fixed operation without learning.

Answer

All three stages are transformers, so none of them learn anything during fit(); they only reshape the data.

Answer

Only LogisticRegression is an estimator; StringIndexer and VectorAssembler are both pure transformers that require no fitting.

Answer

VectorAssembler and LogisticRegression are estimators, while StringIndexer is a transformer applied before assembly.