Back to Projects

ML Competition

Kaggle Fertilizer Competition

A compact competition project where I practiced building a clean modeling loop: inspect the data, prepare features, compare models, generate predictions, and keep the submission process reproducible.

Period

Jun 2025

Stack

PythonPandasScikit-learnFeature Engineering

Motivation

Kaggle is useful because the feedback is concrete. This project helped me practice making modeling decisions under a metric instead of relying on vague impressions.

Links

Core Features

  • Prepared structured competition data for modeling.
  • Experimented with features and model choices.
  • Organized code for repeatable prediction and submission.

My Contribution

  • Handled preprocessing, feature construction, and model experimentation.
  • Used the project to strengthen end-to-end competition workflow discipline.

What I Learned

  • Improved comfort with evaluation-driven iteration.
  • Learned to balance model complexity with a workflow that stays easy to rerun.