Inspiration
I wanted to create a ranking for teams based on specific stats each team has. I gathered each player's stats individually since some players swap teams after a certain time.
What it does
Calculate teams based on stats and averages it out which helps compare each team to one another.
How we built it
The backend and grab data files goes through each game and grabs data at the 10 minute and 20 minute marks, and which team won at the end and how long it took to win or lose. The values are multipled by a weight calculated through linear regression based on lcs summer 2022.
Challenges we ran into
Due to time constrains, I could not test thoroughly and could not create the frontend as I wanted with checkboxes of teams and tournaments to make it easier to use the API. I also realized I did not have any indicator for tournament tiers so collegiate teams and world teams were compared similiarly. The virtual machine with the free tier was low on memory, so to fix this I chunked the json data by reading character by character until I found the object then saved it. This was slow, so I used the free credits to upgrade the VM, but I still have the solution ready in case I need to downgrade the VM again. Lastly, the data for players, teams, and other indicators were not always present so it made it required some changes or just to skip over it.
Accomplishments that we're proud of
I am proud that I got to interact with the abundance of data and look through it to find what I needed. I also learned how to use the ML Linear Regression model to let it figure out the weights for me.
What we learned
I learned that it is very hard to figure out the correct features to find which team is the best and finding those stats for the features is also a difficult task.
What's next for League Team Rankings
I would like to fix up the frontend to make it more user friendly and easier to view the results. Also I would like to add another feature that indicates the tournament tier which will further separate the team rankings. The missing data was also a problem I would like to fix, but would be last on the list of to-dos.
Log in or sign up for Devpost to join the conversation.