A sports analytics pipeline engineered to look beyond standard points tables and mathematically measure spectator value ("entertainment index") across 8 seasons of the Zimbabwe Premier Soccer League (ZimPSL).
- Automated Data Scraping: Built custom Python extractors parsing 101 team-season records from historical league tables (2017–2024).
- Feature Engineering: Standardized goal rates, match volatility, and defensive risk metrics into a unified Chaos & Entertainment Index.
- Tactical Profiling: Mapped team playstyles into tactical quadrant matrices to isolate high-stakes, entertaining sides from low-volatility defensive teams.
Python • BeautifulSoup • Pandas • NumPy • Matplotlib / Seaborn
📊 Read Full Analysis Report | 💻 View Scraper Code
An institutional machine learning application designed to resolve pricing opacity and asymmetry in the Zimbabwean secondary automobile market. The system automatically scrapes marketplace listings, models non-linear vehicle depreciation with target log scaling, and serves interactive bargaining guidance.
-
Variance Explained (
$R^2$ ): 0.790 (~79% of cross-market price variance explained) - Predictive Accuracy (MAE): ~$5,112 USD average baseline dollar error across all vehicle segments
- Dataset Scope: 3,000+ cleaned vehicle records scraped across regional classifieds (ZimAuto, ZimClassifieds)
-
Statistical Governance: Real-time residual Z-score filtering (
$\pm3\sigma$ ) to catch data anomalies and structural risks
Python • BeautifulSoup • XGBoost • Optuna • Scikit-Learn • Gradio • Render
👉 Read Full Analysis | Live Demo