seokhyeonsong / lol_winning_team_prediction_data_science

This is repository for data science team project

LOL_Winning_team_prediction_data_science

This is repository for data science term project.

Summary

There are 3 kinds of codes

1. data plotting
2. real-time prediction
3. model analysis

Data plotting code will show the analysis of our data
Real-time prediction code will receive event from you, and will predict whether win or lose
Model analysis will show the analysis of our used learning model - kNN, k-means

Data plotting

Data plotting will show how data is consisted

data_explore.py will work for data plotting

Data is from here with 10.13 patch version, Korea server and kaggle data
First data is used for prediction result is same. To be referred.
Second data is used for prediction.
If you run data plotting code, you may see the various analysis of our data

Real-time prediction

main.py will work for real-time prediction and model analysis

If you run real-time prediction, you may get the prediction of your game.
Each input should be done by hand with keyboard.
Prediction is used with knn algorithm - which we implemented not module.
If prediction result is same as win-win or lose-lose, then average winning rate of your team will decides the win

Model analysis

 dirty_processing_main.py will work for real-time prediction and model analysis

We analyzed our data with knn and k-means
knn will show confusion matrix k-fold evaluation with hyper-parameter tuning
k-means used ensemble learning to enhance algorithm and show confusion matrix of our data