Introduction

Football has been the most-watched sport all over the world. Every matchday people come up with different opinions on which team wins, loses, or draws. Now with the help of AI, we can leverage a football match outcome based on the team's performance, and stats with the help of the MindsDB Machine Learning (ML) model.

Here is a live app to predict English Premier League scores.

What is Mindsdb?

MindsDB is an open-source tool that allows you to create predictive models using a dataset of your choice. With MindsDB, you have the ability to build, train, optimize, and deploy your Machine Learning models without the need for other platforms.

In this tutorial, we will explore how to use MindsDB to predict the football match outcome for English Premier League.

The dataset used to forecast match Outcomes

The dataset used was gotten from Kaggle and can be found here. It contains 20 years of EPL matches dataset and is presented in CSV format.

The English Premier League is the most popular domestic team in the world. It includes information about the match date, time, home and away teams, and detailed match statistics

Connecting the Dataset with Supabase

Getting started with Mindsdb

MindsDB provides all users with a free MindsDB Cloud version that they can access to generate predictions on their database. You can sign up for the free MindsDB Cloud Version by following the setup guide. Verify your email and log into your account and you are ready to go. Once done, you should be seeing a page like this :

Integrating MindsDB with Supabase

MindsDB provides us the ability to integrate with Supabase. Supabase is an open-source Postgres database that allows developers to store and sync data across multiple devices in real-time, a Firebase alternative.

Importing Our Dataset to Supabase

Head to the Supabase dashboard and create a new project if you can sign up, create a new project, follow the prompt, and enter your organization name, project name, and password.

​Note: copy and save the password somewhere for later.

click on Create new project then wait for the project to finish setup

Click on the table editor icon on the left sidebar then create a new table, you can create a table called "epldata"

Click on import data via spreadsheet and import the CSV datasets you downloaded from Kaggle. We will be working with the 2021-2022 data file

Note: Before importing the dataset to Supabase, edit the dataset on Microsoft Excel or any supported software. change the A column Div to id respectively to make the id column a primary key as seen in the image below:

Click on the settings on the left sidebar then click on Database for the information needed to connect the Supabase with mindsDB

Head back to the MindsDB console and run the following SQL command. Replace Parameters data respectively with your Supabase database information above.

CREATE DATABASE supabase_datasource  --- display name for the database
WITH ENGINE = 'supabase',  --- name of the MindsDB handler
PARAMETERS = {
  "host": "127.0.0.1",  --- supabase database host
  "port": 54321,  --- supabase database port
  "database": "test",  --- supabase database name
  "user": "supabase", --- supabase database user
  "password": "password"  --- supabase database password
};

Understanding the Dataset

Our main goal is to predict the Full Time Result (FTR) of the match. Our database consists of various fields but the important Columns are listed below :

• Date = Match Date (dd/mm/yy)

• HomeTeam = Home Team

• AwayTeam = Away Team

• FTHG = Full-Time Home Team Goals

• FTAG = Full-Time Away Team Goals

• FTR = Full Time Result (H=Home Win, D=Draw, A=Away Win)

• HTHG = Half-Time Home Team Goals

• HTAG = Half-Time Away Team Goals

• HTR = Half-Time Result (H=Home Win, D=Draw, A=Away Win)

• Referee = Match Referee

• HS = Home Team Shots

• AS = Away Team Shots

• HST = Home Team Shots on Target

• AST = Away Team Shots on Target

• HHW = Home Team Hit Woodwork

• AHW = Away Team Hit Woodwork

• HC = Home Team Corners

• AC = Away Team Corners

• HF = Home Team Fouls Committed

• AF = Away Team Fouls Committed

• HO = Home Team Offsides

• AO = Away Team Offsides

• HY = Home Team Yellow Cards

• AY = Away Team Yellow Cards

• HR = Home Team Red Cards

• AR = Away Team Red Cards

we can run the following query in the MindsDB Console to preview the data we will use to train our model:

SELECT * 
FROM supabase_datasource.epldata  
LIMIT 10;

Creating and Training our Prediction Model

we can run the following query in the MindsDB Console to train our model to predict the football match outcome

CREATE MODEL 
  mindsdb.predict_scores_model
FROM supabase_datasource
  (SELECT * FROM epldata)
PREDICT FTR 
GROUP BY FTHG, FTAG;

Here predict_scores_model is the name of our Predictor Model and FTR is the target value that we want to predict. Click on Run or press Shift+Enter to run our query. If we don't encounter any issues, we will get a Query Successfully Completed message.

Checking the Status of our model

Here we check the status of the newly created model we run the query in the MindsDB console as seen below :

SELECT * 
FROM mindsdb.models 
WHERE name='predict_scores_model';

Making Predictions to Our Model

Once the model's status is complete. We can now make predictions of our model using the SELECT statement to query for prediction results.

We can now query our model to predict football match outcomes by running the following query in the MindsDB console as seen below with the output :

SELECT m.HomeTeam, m.AwayTeam, m.FTHG, m.HTHG,m.FTR as winner, FTR_explain
  FROM mindsdb.predict_scores_model as m 
  JOIN supabase_datasource.epldata as t
  WHERE t.HomeTeam = "Arsenal"
  AND t.AwayTeam = "Burnley";

We also query our model to predict football match outcomes in batch by running the following query in the MindsDB Console as seen below with the output :

SELECT * 
FROM supabase_datasource.epldata
JOIN mindsdb.predict_scores_model as m limit 10;

Conclusion

You have successfully created a predictive model using MindsDB and Supabase. You can now use this model to make predictions on new data. You can also experiment with different predictor types and configurations to see what works best for your data.

You can also clone the GitHub repo to run the app on your local machine.

If you find this tutorial helpful let me know in the comment section and don't forget to like and share.