Steam Games Analytics

Steam Games Analytics starts with more than 124,000 games from the FronkonGames dataset on Hugging Face. I load the data into DuckDB, clean it with dbt, and turn it into tables that are easier to query and use in a dashboard.

Overview

The models move through staging, intermediate, and marts layers. Along the way they standardize fields, remove duplicates, flatten genres and tags, calculate review sentiment and engagement, and estimate revenue from owner ranges and prices.

What I worked on

The marts cover game performance, genres, publishers, and engagement segments. The project also has 24 data tests for things such as non-negative review counts, valid prices, logical release dates, uniqueness, and accepted score tiers.

The dashboard is available at ilhamradhan.github.io/steam-games-analytics.