ML Project: Predicting Football Match Outcomes

Data Preparation

Loading & Cleaning: Read a CSV of 31,500+ club matches (July 6, 2019 – June 1, 2025) with low_memory=False and dropped any rows missing values.

Encoding: Converted every text column (team names, league, dates, and the result label H/D/A) into numeric codes so the model could understand them.

Splitting: Separated the FTResult column (our target) from all other features, then split into 80% training and 20% testing sets, stratified by result.

🏋️‍♂️ Modeling

Algorithm: Random Forest classifier with 200 trees and a maximum depth of 10.

Training: Fitted the model on the training set until it learned patterns in possession, form, and odds.

Performance: Achieved 98% accuracy on the unseen test matches—far beyond my initial expectations!

Visualization

Plotted the first 100 actual vs. predicted FTResult values, marking each point by H (home win), D (draw), or A (away win). Watching the markers overlap so closely was truly exhilarating.