Football or Soccer(as they call it in America) is a game that is loved by millions across the world. One day, while checking out the player performances, it struck me that the **Player Ratings **could be **predicted **with the help of Machine Learning techniques. So, I checked out for public datasets but couldn’t find the kind of data that I was looking for. So, I did some web-scraping and gathered data for the season 2019–20 from various leagues like Barclays Premier League, La-Liga, Bundesliga, Serie-A, Champions League etc. It is available on Kaggle (https://www.kaggle.com/sanjit1105/soccer-players-stats-and-ratings-matchwise). So, let’s get started!

Description of the features

The Dataset has 47 features and our Target Variable is “Rating” which ranges from 0 to 10. If we take a look at our dataset features, we observe that most of the features are numerical features where an increase or decrease in the value can substantially affect the outcome or prediction. We will scale some of the values but some values are not exactly scalable so we will leave them as it is. There are categorical features such as is_a_sub, was_subbed, yellow_card and red_card.

Data Exploration

Let’s start by importing the .csv file and libraries such as pandas,seaborn, **matplotlib **and **numpy **as they might be useful later on.

#football-predictions #machine-learning #soccer-prediction #football #sports