Instead of studying the top games, let’s take a look at the bottom feeders instead!
My data visualization focuses on the least popular scores, tags, and reviews among the top 1,000 games published on Steam up to 2026. The style is purposefully geeky and playful, while remaining minimalistic and slightly serious, since the intended audience is, after all, geeks and gamers.
This analysis serves two purposes: first, to explore unclaimed market opportunities for smaller indie game developers; and second, to help gamers discover lesser-known titles in rare or niche genres.
I believe the final images could be used either as wall posters or as promotional magazine spreads.
First, I found this dataset on kaggle and opened it in Google Sheets to take a closer look at which stats can be analyzed. I’ve realized that this dataset includes some amount of DLCs and bundles and skins — which needed to be removed by hand in order to not skew the tag count.
Then in Visual Studio Code I’ve created a Jupyter Notebook and used pandas to count and split tags… Once again, I had to remove certain titles because they were too similar in genre or contained only a single tag, which would have resulted in relatively uninformative data.
Lastly, I adjusted my plot a little further to make the graphic more visually appealing and to make the stat «pop.»
To discover «hidden gems,» I created a correlation between high review scores and low ownership counts. This was absolutely beyond my pandas and matplotlib knowledge, so I asked ChatGPT to help me vibecode it together.
Personally, I don’t know much math at all, so I can only put my trust in machine intelligence and hope for the best.
The final result seems pretty decent, and I was even surprised to find some titles I’ve played myself and can genuinely vouch for. «The Liar Princess and the Blind Prince,» for example, is a fantastic game, and I don’t know a single other person who has ever played it.
The final graph also relies on correlation and AI help. I’m dividing the data into four year-based groups to see whether the release year has an effect on review scores. My original theory was that people tend to give higher scores to older games, since we humans do have a tendency to romanticize older things.
Theory was proven correct—up until the very last year group, where review scores suddenly spike again. I assume that’s because new releases are generally more exciting to players for a while. They may also benefit from paid critic reviews and increased media attention aimed at boosting launch sales.
My data source: kaggle dataset Top 1000 Steam Games (2024–2026) by Waddah Ali
Mockups used: Alexandre Lallemand, DVLOOP CREATIVES, Graphic Time
Software used: ChatGPT, Google sheets, Visual Studio Code, Clip Studio Paint EX, Adobe Photoshop
