user@devops:~$ cat README.md
Pipelines — Preprocessing with Titanic
# Description
scikit-learn Pipelines project using the Titanic dataset. Learn to encapsulate all preprocessing in a single Pipeline object: null imputation with SimpleImputer (median for numeric, mode for categorical), scaling with StandardScaler, category encoding with OneHotEncoder, and combination with ColumnTransformer. The complete pipeline is passed to GridSearchCV to optimize both preprocessor and model parameters simultaneously. Compare with the manual approach to see why pipelines prevent the most common error: data leakage.
# Key features
$ Pipeline: complete sequence of transformations + model in one object
$ ColumnTransformer: different transformations for each column type
$ SimpleImputer: automatic null value imputation
$ OneHotEncoder: categorical variable encoding
$ StandardScaler: numeric variable scaling
$ GridSearchCV with Pipeline: simultaneous optimization of the entire flow
$ With/without pipeline comparison: data leakage prevention
# Gallery
# Technologies used
