Applying Machine Learning Techniques to Predict the Outcome of Football Matches

Demo App

-Flask -pandas -numpy -selenium -IPython -sklearn -xgboost -scipy -requests -gunicorn

python3 model.py

• Includes necessary helper functions to process raw data

• This collects and adds more data to the raw data.
• current standings and goals for and against etc.

• This collects the current fixtures and match results.

• This calculates the league points and generates standings

• Scrapes the overall team data from fifa

• With using processed data, train a ML model to predict future results

• I/O file where the functions from the above files are actually executed.

• scraped overall team stat data

• historical standing results calculated in rankings.py

• manually collected historical data of match outcomes
• latest match outcomes of the current season

• data extracted from data/raw

• data processed from data/raw_cleaned

1. data/statistics/round_rankings (directory)
   • standings calculated based on the predicted match outcomes
   • each file in the directory has a date included in its name. It provides predicted standing outcomes at the denoted date
2. data/statistics/prediction_ranking.csv
   • predicted standing at the end of the season
3. data/statistics/prediction_result.csv
   • individual predicted match outcomes
4. data/statistics/round_rankings_summary.csv
   • predicted standing summary over the course of the season

• disk cache of classifier that gives the best accuracy of prediction

• sql database that stores previous match outcomes, predicted match results and predicted standings

• csv file used for training a model and making predictions

• list of grid searched classifiers and its confidence score