Predicting Amazon Electronics Best-Sellers

A CRISP-DM-driven supervised classification project on a 42K+-item Amazon Electronics sales dataset (2025) to predict best-seller status and surface the product-level features most predictive of best-seller likelihood.

Final project for DS1312 — Data Mining, BS Data Science, University of Asia and the Pacific. Group: Riego + Camacho.

Full title: Predicting Amazon Electronics Best-Sellers: Identifying Key Features Through Machine Learning

Repository layout

notebook/      Main project notebook (end-to-end pipeline)
data/
  raw/         Original + cleaned Amazon Electronics datasets (~36-39 MB each)
  ABOUT THE DATASET.pdf
documents/     Final paper, project proposal, earlier iterations, sample PDF

Methodology — CRISP-DM

1. Business understanding — "what makes an Amazon Electronics product a best-seller?"
2. Data understanding — profile 42K+ rows; identify key product features.
3. Data preparation — cleaning, encoding, feature engineering.
4. Modeling — supervised classification (multiple algorithms compared).
5. Evaluation — metrics, feature-importance interpretation.
6. Deployment / Communication — paper + presentation.

Stack

Python · pandas · scikit-learn · matplotlib · seaborn

Status

Course final project, completed Jan–Apr 2026.

---

David Nathaniel P. Riego · BS Data Science, UA&P (Aug 2023 – Aug 2027 expected) · LinkedIn