This repository contains Python code to study the relationship between review numbers and sales for Steam games.
- 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
- 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
- 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
- Steam API's
appreviews
endpoint, to get a dump of reviews on an application in Steam steamreviews
: a PyPI package to download Steam reviews
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'scurve_fit
for 1D signals.lmfit
: a wrapper around SciPy'sleastsq
,- Some documentation by
scipy
andsklearn
:
- 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
AmongUs-DAU
: compute "Daily Active Users" (DAU) of Among Us from "Concurrent Connected Users" (CCU)