The essence of the project is to develop a model that can predict the outcome of a match for two teams, taking into account the following distinctive features:
- teams play only within their own league, but the model should be able to predict including the result for two teams from different leagues
- the matches are divided by time, respectively, the model should give an up-to-date result for teams that are within the same season
- for some teams, the data is incomplete, that is, there are not all statistics for 38 games for the season
The file with the source database could not be uploaded to the repository, but it can be obtained by downloading it from the kaggle website (url: https://www.kaggle.com/code/jacobopedrosa/simple-football-predictions-accurance-75-40/data)
In a text file sql_code.txt contain queries for a combination of tables from the source database. Accordingly , in the file match_info.csv information on the results of each match is generated. The file team_skills.csv displays the attributes of the commands, player_skills.scv - displays the attributes of the players.
In the file Football_matches_results_prediction.ipynb the code for solving the problem is given. The stages of work, possible models, results and comments can be found there