ocansey11 / betting_prediction_model

Machine Learning Model for Bet Tipping sites. By scrapping Data we are creating a model to measure credibility for bet tips and odds

Betting Prediction Model

This project aims to predict football match outcomes using data scraped from various sources (Forebet.com and Sofascore). The project includes data scraping, processing, exploratory data analysis, model training, and evaluation.

Project Structure

Data

• data/csv/: Contains the weekly accumulated and updated training(completed_match + previous league stats) and testing datasets(upcoming_matches + current league stats).

Notebooks found in various directories

• notebooks/data_scraping/: Notebooks for scraping completed matches, upcoming matches, and league table standings.
• notebooks/data_processing/: Notebooks for combining the scraped data into training and testing datasets.
• notebook/exploratory_data_analysis.ipynb: Notebook for exploring the data. finding interesting patterns as the data grows weekly
• notebook/model_classifiers.ipynb: Notebook for training and comparing different classifiers. The top three classifiers from this notebook will be expanded on. As the data grows new classifiers might perform better than old ones.
• notebooks/models: Notebooks for tuning the best models and testing them out.

SQL Scripts

• sql_scripts/: SQL scripts for creating tables and inserting data. incase you want to try and edit somethings on your own

Source Code

• modules/: Python modules for database connections and SQL operations.

Flowcharts

• flowcharts/: Diagrams explaining the data flow and processing steps.

How to Use

Option 1: Follow the Full Process

1. Scrape Data: Use the notebooks in betting_prediction_model/data_scraping/ to scrape the latest data hence creating and storing them in a db
2. Process Data: Use the notebooks in betting_prediction_model/data_processing/ to combine the scraped data into training and testing datasets.
3. Train Models: Use betting_prediction_model/comparing_classifiers.ipynb to train and compare different models. and pick which one fits best. Currently we have little data. When the next season starts we will have really good data to work on
4. Tune Models: Use betting_prediction_model/models to tune the best models individual. This helps you focus on your favorite model and play around with parameters and hyperparameters to yeild the best results

Option 2: Use Pre-Processed Data

1. Download Data: Download the weekly accumulated and updated training and testing datasets from betting_prediction_model/data.

2. Train Models: Use betting_prediction_model/comparing_classifiers.ipynb to train and compare different models using the downloaded datasets.

3. Tune Models: Use betting_prediction_model/models.ipynb to tune the best models.

4. Automation: You can run Pulling training and test csv from 'data' to directly pull training_data.csv and testing_data.csv into your database. This way you run the models straight away after.

Requirements for Automation and others

ChromeDriver My sql Workbench Python 3.0+