woctezuma / steam-reviews-to-sales

Study the relationship between review numbers and sales for Steam games

Steam reviews to sales

This repository contains Python code to study the relationship between review numbers and sales for Steam games.

Requirements

• Install the latest version of Python 3.X.
• Install the required packages:

pip install -r requirements.txt
pip install git+https://github.com/google/pwlfit.git

Usage

• To download review numbers as they were on July 1, 2018 for 13,275 games of interest, run:

python download_data.py

Data is downloaded to data/review_stats.json.

• To filter extreme values and display a scatter plot of the data, run:

python analyze_data.py

• To fit a model to the data, run:

python fit_model.py

References

Data leak for sales

• Ars Technica's article: Valve leaks Steam game player counts; we have the numbers, July 2018
• Ars Technica's supplemental data: snapshot of "players estimate" for 13,281 games

Data download for reviews

• Steam API's appreviews endpoint, to get a dump of reviews on an application in Steam
• steamreviews: a PyPI package to download Steam reviews

Regression tools and documentation

• mapie: Model-Agnostic Prediction Interval Estimator (MAPIE),
• pwlfit: Google's library to fit piece-wise linear functions,
• statsmodels: econometric and statistical modeling,
• tidfit: a wrapper around SciPy's curve_fit for 1D signals.
• lmfit: a wrapper around SciPy's leastsq,
• Some documentation by scipy and sklearn:
  • supervised learning
  • common pitfalls with linear regression
  • losses for robust fitting

Blog posts

• Jake Birkett, How to estimate how many sales a Steam game has made, March 2015
• Jake Birkett, Using Steam reviews to estimate sales, May 2018
• Simon Carless, How that game sold on Steam, using the 'NB number', August 2020
• VG Insights, How to Estimate Steam Video Game Sales?, August 2021
• Simon Carless, Epic advances, Steam reviews, oh my!, August 2021
• VG Insights, Further analysis into Steam Reviews to Sales ratio, October 2021
• VG Insights, Steam sales estimation methodology and accuracy, June 2022
• Gamalytic, How to accurately estimate Steam game sales, May 2023
• Gamalytic, What makes people review your game? A deep dive into the Steam's sales/review ratio, July 2023
• Simon Carless, What 'Steam review count' tells us about your game, August 2023

Other repositories of mine

• AmongUs-DAU: compute "Daily Active Users" (DAU) of Among Us from "Concurrent Connected Users" (CCU)