🎮 Video Game Rating Analysis

A Data Science project that analyzes Metacritic video game review data and builds machine learning models to predict critic scores.

Author: Mushtaque Ali | CMS-ID: 023-23-0165 | Course: Data Science

---

📋 Project Overview

The video game industry is one of the fastest-growing entertainment sectors worldwide. This project applies the complete Data Science lifecycle to analyze video game review data and build a predictive model that estimates Metacritic scores based on available features.

Objectives

• Understand trends in video game ratings
• Explore relationships between critic scores and user reviews
• Build a regression model to predict Metacritic scores
• Evaluate and compare model performance

---

📁 Dataset

Property	Detail
Source	Metacritic Video Game Reviews (Public Dataset)
Rows	18,800
Columns	6
Time Period	1995 – 2021
Platforms	22

Features

Column	Description
name	Name of the video game
platform	Gaming platform (PS, Xbox, PC, etc.)
release_date	Release date of the game
summary	Short description of the game
meta_score	Metacritic critic score (Target variable)
user_review	Average user review score

---

🛠️ Tech Stack

• Language: Python
• Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn
• Notebook: Jupyter Notebook
• Models: Linear Regression, Ridge Regression, Random Forest, Gradient Boosting

---

🔄 Project Pipeline

Data Collection → Data Cleaning → EDA → Feature Engineering → Model Building → Evaluation

Data Cleaning Steps

• Removed missing values from meta_score and user_review
• Converted user_review to numeric datatype
• Removed non-numeric / invalid entries
• Encoded platform using Label Encoding
• Extracted year and month from release_date
• Dropped unnecessary text columns for modeling

---

📊 Key Findings

• Average Meta Score: 70.7 | Median: 72.0
• Average User Review: 6.99 | Median: 7.30
• Critic-User Correlation: 0.526
• Most games score between 60–80 on Metacritic
• Both critic and user scores show a declining trend over time
• PC has the most games (4,829), followed by PlayStation 4 (2,039)

🏆 Top Platforms by Average Meta Score

Rank	Platform	Avg Meta Score
1	Nintendo 64	78.4
2	Xbox Series X	76.0
3	PlayStation 5	75.4

---

🤖 Model Results

Model	R² Score	MAE	MSE
Linear Regression	—	—	—
Ridge Regression	—	—	—
Random Forest	—	—	—
Gradient Boosting ✅	0.378	0.789	—

Best Model: Gradient Boosting with R² = 0.3780 and MAE = 0.7885

Features Used for Modeling

• Platform (encoded)
• User Review Score
• Release Year
• Release Month
• Summary Length

---

⚠️ Limitations

• Game summary text was not used (NLP could improve results)
• Non-linear relationships may not be fully captured
• External factors like marketing and franchise popularity were not included

---

🚀 Future Work

• Apply NLP / Sentiment Analysis on game summaries
• Use advanced models like XGBoost or Neural Networks
• Include additional features (genre, developer, publisher)

---

📂 Project Structure

├── all_games.csv               # Raw dataset
├── Project.ipynb               # Main Jupyter Notebook
├── Report_VIDEO_GAME_RATING_ANALYSIS.pdf  # Project report
└── README.md                   # This file

---

📜 References

1. Metacritic Video Game Dataset
2. Scikit-learn Documentation
3. Pandas Documentation
4. Matplotlib Documentation
5. Python Official Documentation