Beta*
A project to pit predictions to reality using machine learning;
A repo to predict soccer scores from Historical train data and timetable &/test data.
We use Scikit-Learn's RandomrestRegressor to predict soccer scores purely based on historical data and betting odds. https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html
By data enthusiasts for betting enthusiasts.
Steps to follow:
1. Run Step-1-To download Historical Data (1993-2019).py to download historical data as available on http://football-data.co.uk/ on major Eurpean Leagues and save it into one consolidated csv which will be used as training data
2. Run Step-2-Download Updated Data and concat with Historical Data.py to update training data to reflect latest results
3. Run Step-3-BET365 odds scraper_raw.py to download https://www.bet365.com odds for all respective leagues. This is required to predict outcomes. We use https://github.com/S1M0N38/soccerapi to scrape betting odds.
4. Run Step-4-Normalizing Odds.py to normalise team names and split odds dict list into a flat consumable format which will be used as testing data
5. Run Step-5-Predicting.py to predict outcomes present in testing data