fifa

FIFA Player Value – Machine Learning Project

📌 Project Overview

This project analyzes FIFA player data and applies Machine Learning techniques to predict and classify player value. The work includes data exploration, feature engineering, clustering, regression, and classification models.

🎯 Goal

The original goal was to predict a player's market value using regression models. Later, the problem was reframed as a classification task, categorizing players into value groups (low / mid / high).

---

📊 Dataset

• Source: FIFA players dataset
• Rows represent players
• Features include player attributes, skills, and statistics

---

🔍 Exploratory Data Analysis (EDA)

• Distribution analysis of numerical features
• Correlation analysis
• Log transformation applied to the target variable to handle skewness

---

🛠 Feature Engineering

• Scaling of numerical features
• Encoding of categorical variables
• Creation of new meaningful features

---

🧩 Clustering

• Applied K-Means clustering
• Cluster labels and distance to centroid were added as features
• Helped identify natural groups of similar players

---

🔄 Regression to Classification

• Converted the continuous target into discrete classes using quantile binning
• Created 3 classes:
  • Low value
  • Medium value
  • High value

---

🤖 Classification Models

The following classification models were trained and evaluated:

• Logistic Regression
• Random Forest
• Gradient Boosting

Evaluation Metrics:

• Precision
• Recall
• F1-score
• Confusion Matrix

Recall was considered more important than precision, especially for identifying high-value players.

---

🏆 Best Model

• Random Forest Classifier
• Achieved the best balance between recall and precision
• Strong performance on high-value player classification

---

💾 Model Export

The winning model was exported to a pickle file and uploaded to this repository.

📁 File:

• random_forest_fifa.pkl

---

🎥 Presentation Video

📺 Link to project presentation video:
[PUT YOUR VIDEO LINK HERE]

---

✅ Summary

This project demonstrates an end-to-end ML pipeline: EDA → Feature Engineering → Clustering → Modeling → Evaluation → Deployment.