Free Steam Game Sentiment Analysis

The goal of this project is to analyze two free games on Steam and take a look at how they are perceived using R. For this project, I took a look at CS:GO and Dota 2, both games which are owned by Valve, the company that runs Steam. I took 100 reviews of each of these games right from Steam’s website with the help of their steam review API. After gathering these reviews, I changed tokenized them by word using the tidytext package.

Word Usage

For the first bit of analysis, I wanted to take a look at the most commonly used words in the reviews for these games. I wanted to get an idea for common trends in these games.

Unsurprisingly the most commonly used words, when taking out most stop words, are still mostly neutral. The words “game”, “dota (2)”, and “playing” all just have to do with describing what the individual is reviewing, but the rest of the words, most notably “hackers”, “smurfs”, "f**k“,”bad“, and”fun" can begin giving an idea at the perception of these games. From what we see here so far, it seems that CSGO may have a much more negative perception than Dota 2.

Sentiment Analysis

By breaking down these reviews by word, I was able to run a sentiment analysis on them.

Just by looking at a pure total’s perspective, it would seem that Dota 2 reviewers write many more non-neutral words than CSGO reviewers do in their reviews. With that said, it seems that both CSGO and Dota 2 have proportionally similar positive to negative sentiments in their reviews. Dota 2 may be slightly more positive, but overall they are similar. Fear and sadness seem to be a more common trend in these games than joy, which for a game you would hope to not be the case, I would imagine. Anger seems to be much more prevalent in Dota 2 than it does in CSGO, however Dota 2 players seem to have more anticipation.

Recommendations over the Year

The data I collected did not lend itself for useful comparisons, and with the time I have to complete this and the rest of what I have this week I’m just going to take the L on this section, and improve on it for the final project. I was wanting to animate the data by year, but I realized that the data for each game were only from 2021 for CSGO and 2017 for Dota 2. I tried fixing this but ended up breaking too many other things. The goal was to show reviews for CSGO and Dota 2 changed (recommended or not) over the years 2017-2021.