An estimator is the final pipeline component: it trains a predictive model from transformed input data and target values. section 3.4.3
Within a scikit-learn workflow, an estimator:
- Receives features produced by preprocessing transformers. section 3.4.3
- Is refit inside each cross-validation fold when contained in a pipeline. section 3.4.3
Estimator in a molecular-classification pipeline
A pipeline standardizes SMILES, generates molecular fingerprints, and passes them to SGDClassifier, the estimator. GridSearchCV evaluates parameter combinations using five-fold cross-validation and F1 scoring.




