What characterizes a Spark ML estimator?

A Spark ML estimator is characterized by its ability to be fit on a DataFrame to produce a Transformer. Estimators in the Spark ML library implement algorithms for learning from data. When an estimator is trained (or fit) on a dataset, it computes the necessary parameters to create a model, represented in the form of a Transformer. The resulting Transformer can then be utilized to transform the input data into predictions or other derived features.

This distinction is fundamental in the Spark ML workflow, where the separation between estimators and transformers allows for a clear delineation of the training process (where estimators are involved) and the application of learned models (via transformers). Thus, the core function of an estimator is to define the learning algorithm that will be executed to derive the model parameters.