Metacritic Game Ratings Analysis with PySpark

This project was completed as part of a university assignment focused on data analysis using PySpark and received a final mark of 71%. The task involved querying and summarising a large dataset of Metacritic game reviews to uncover insights about platforms, genres, and developers.

Objectives

• Filter and summarise games with high critic scores (e.g. ≥ 95)
• Group by developer to calculate:
  • Number of titles
  • Average critic/user scores
  • Total critics/users
• Identify the developer with the most multi-platform releases
• Detect the platform with the highest share of open-world games
• Count distinct single-player titles and their associated developers

Tools and Technologies

• PySpark (DataFrames API)
• Google Colab
• CSV dataset (~10,000+ rows)

Skills Demonstrated

• Data wrangling and transformation with PySpark
• Filtering, aggregation, and grouping
• Handling multi-platform duplicates
• Applying logical conditions to string fields (e.g. identifying “Single Player” titles)
• Working with large datasets in a distributed computing environment

Notes

This assignment was graded via CodeGrade with AutoTests and received a 71%.
Although not all outputs were exact matches, all logic and core query structures were successfully implemented.

Example Queries

Below are representative PySpark queries used in this project. These showcase key data transformations and aggregations applied to the Metacritic dataset.

# 1. Filter games with score ≥ 95 released after 2017
top_games = df.filter((df['score'] >= 95) & (df['release_year'] >= 2017))
top_games.select('title', 'score', 'release_year').show()

# 2. Developer summary: number of titles, average scores, totals
from pyspark.sql.functions import avg, countDistinct, sum

developer_summary = df.groupBy('developer').agg(
    countDistinct('title').alias('num_titles'),
    avg('score').alias('avg_critic_score'),
    avg('user_score').alias('avg_user_score'),
    sum('num_critics').alias('total_critics'),
    sum('num_users').alias('total_users')
).orderBy('num_titles', ascending=False)
developer_summary.show()

# 3. Identify multi-platform titles (appear on ≥ 2 platforms)
multi_platform = df.groupBy('title').agg(
    countDistinct('platform').alias('platform_count')
).filter("platform_count >= 2")
multi_platform.show()

# 4. Developer with most distinct platform releases
most_platforms = df.groupBy('developer').agg(
    countDistinct('platform').alias('distinct_platforms')
).orderBy('distinct_platforms', ascending=False).limit(1)
most_platforms.show()

# 5. Platform with highest percentage of Open-World titles
from pyspark.sql.functions import col

open_world = df.filter(df['genre'].contains('Open-World'))
open_world_count = open_world.groupBy('platform').count().alias('open_world_count')
total_count = df.groupBy('platform').count().alias('total_count')

platform_percent = open_world_count.join(total_count, 'platform') \
    .withColumn('percent_open_world', (col('open_world_count') / col('count')) * 100) \
    .orderBy('percent_open_world', ascending=False)
platform_percent.select('platform', 'percent_open_world').show()

# 6. Count distinct Single Player titles and developers
single_player = df.filter(
    (df['feature_player'] == '1 Player') | (df['feature_player'] == 'No Online Multiplayer')
)
num_titles = single_player.select('title').distinct().count()
num_developers = single_player.select('developer').distinct().count()