Today I Found: Predicting Success on Steam

Today's recommendation of an article I found is "Summary: Predicting Success of Steam Games" by Michal Trněný using machine learning to predict a game's success before its released by using various features like tags, the launch price and... average saturation and number of distinct colors of the first displayed screenshot...? Sounds interesting, so please have a look! Interesting observation regarding a game's tags:

Those that stand out as unsuccessful include: 2D, pixel graphics, platformer, point & click, puzzle, retro. Games with these tags are generally doing pretty poorly. [...] Games with the following tags have been mostly successful: third-person (shooter), first-person (shooter), open world, sandbox, survival, story-rich, fantasy, sci-fi, zombies.

'Today I Found' is a little series of games, pictures, articles or anything else which I think other game developers (or, to some extend, other gamers) could find interesting or inspirational. If there's any relation to the shown thing or its creators, I will state it right here.

Best regards,
Steven