Back to Projects

Machine Learning

Adult Census Preprocessing Pipeline

A supervised learning pipeline project on the Adult Census dataset, focused on feature type separation, imputation, encoding, scaling, leakage prevention, and baseline comparison.

Period

May 2026

Tools / Tech

PythonScikit-learnColumnTransformerPipelineDecision TreeOne-hot Encoding

Why I built it

I used this project to practice the less glamorous but important parts of ML: preprocessing discipline, pipeline structure, and fair evaluation.

Links

What it includes

  • Separates numerical and categorical features.
  • Uses imputation, one-hot encoding, scaling, and model steps inside a pipeline.
  • Compares baselines while reducing leakage risk.

What I worked on

  • Designed preprocessing steps with ColumnTransformer and Pipeline.
  • Implemented baseline classification with attention to train/test separation.
  • Completed the notebook workflow myself and checked how preprocessing choices affect model evaluation.

What I Learned

  • Learned why pipelines are safer than ad hoc preprocessing.
  • Practiced thinking about leakage before model scores.
  • Practiced a more organized ML experimentation workflow.