Apex Legends Stats to Rank Statistical Relationships
- Introduction
My friend told me to play a game called Apex Legends with him when I have some time to play. I don’t really have any time to play games at all anymore, but in the spirit of competition I was thinking about looking and using statistics in the games population to see if I could narrow my focus on the most important things to do on a team to help them win games.
2. Description of the data
I started hunting for the data and couldn’t find any csv or file I could just read into my Data Frame, so I taught myself how to scrape the data I needed from a table on a website. It contained Kills, Damage, Matches Played, Headshots, along with a lot of data I thought I could use to dive into this topic and find some relationship to some dependent variables and Rank.
I found a lot of limitations after I was deep into the rabbit hole, the biggest one was that apparently most of the data comes from if a player decides to use a badge that displays the data they have for 1 out of 10 characters. It means that things every stat outside of total kills, total damage, and Rank Score become instantly useless to me. Not only that but getting info for per match statistics was now impossible.
3. Statistical methods
I decided to continue with this data since I had already spent time on it and I could still use this information to conclude some tests. The first Hypothesis I wanted to test was if there was a relationship between Kills and Rank.
Ho: There is no relationship between Kills and Your Rank.
Ha: There is a relationship between Kills and Your Rank.
I dropped all irrelevant columns, and plotted some graphs to see what I was dealing with. Another limitation showed itself in the ranking system seeming to group large amounts of data at the bottom of each rank, specifically 10,000, 7,200, and 4,800. I decided to try to get rid of those lines the best I could but a lot of people hover around those lines even if it’s not exact.
4. Results
The results of the test were interesting, The results show that the predictability of the line fitted to the graph is pretty poor, but the pstat was low enough to conclude that there is enough evidence to reject the null hypothesis and support the alternative that there is a relationship between kills and your rank.
I decided to also find the relationship between Damage and kills just cause it’s so nice and Linear.
The pstat is low and we have enough evidence to reject the null and conclude there is a statistical relationship between damage and kills. It is also a positive relationship and is fairly predictable looking at the r-squared value of .762.
5. Conclusion
I really wanted to dive deeper, but the data is hard to find from just a website point of view. The stats are kept somewhere but I just couldn’t get access to all the information I needed to dive deep enough to find what I should focus on to help my friend.