FIFA Player Analytics & Performance Prediction

Project DescriptionA data science project focused on end-to-end processing and predictive modeling of FIFA player statistics. The objective was to clean raw player attributes, analyze growth factors, and build accurate machine learning models to forecast player ratings and market valuations.Tech StackLanguages & Libraries: Python, Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn.Key ContributionsData Engineering & Cleaning: Structured messy FIFA datasets by handling null values, removing anomalies, and transforming categorical traits (e.g., foot preference, work rates) into model-ready numerical data.Exploratory Analysis (EDA): Discovered and visualized non-linear correlations between player age, wage scales, and potential using Seaborn heatmaps and distribution plots.Feature Selection: Filtered out redundant attributes using correlation matrices to minimize multi-collinearity, optimizing training efficiency.Predictive Modeling: Built and tuned regression models (including Random Forest and Linear Regression) to predict player $Overall$ ratings and financial valuations.Model Validation: Evaluated performance metrics using $MAE$ and $R^2$ scores to benchmark model accuracy and prevent overfitting.Key OutcomesDelivered a clean, reproducible Python pipeline for parsing and preparing sports player datasets.Quantified the exact weight of individual attributes (like stamina, vision, and pace) on a player's final market valuation.Optimized model hyper-parameters to achieve a high $R^2$ accuracy score in predicting player potential.