FIFA 19 Dataset Analysis

Introduction

FIFA 19 is a football simulation video game developed by EA Vancouver as part of Electronic Arts’ FIFA series. It is the 26th installment in the FIFA series, and was released on 28 September 2018 for PlayStation 3, PlayStation 4, Xbox 360, Xbox One, Nintendo Switch, and Microsoft Windows

Dataset has info about all players that game: 18207 rows and 88 variables.

As a teenager I played a lot in FIFA 09 and had a huge handwritten notebook with information about all potential players, because I always prefered to manage teams (search for talents, reorganise teams) than to play.

As a retake project I need to create Shiny App & Markdown with insights about that dataset.

This reports includes vizualisations and comments.

0. Libraries and reading of data

Coding in R always involves lots of different packages as this very important sign %<>%, to perform operations with dataset, is in one package but parser for dates is in the other, because usual preinstalled parser doesn’t always work.

Data has UTF-8 encoding, if not to specify that, some of the letters or signs might be lost. The first row is dropped because ID is unique, so there is no need in 2 unique values.

1. Data cleaning & feature engineering

As mention above dataset contains: 18297 columns & 88 variables.

From Name to Acceleration of a player, from Favorite foot to Release Clause.

Not all of them are interesting as for example links to photos of players or club logos give mistake 404. I decided to drop 5 columns:

columns_to_drop <- c('Flag', 'Real.Face', 'Height', 'Weight', 'Club.Logo')

There are also other different problems with data such as: - Numeric columns are not numeric because of characters inside - Dates are not seen as dates - NAs

Data was preprocessed and cleaned, new features were created: - Years in club - Height and Weight in KG & CM - Easy to understand positions on the field, as original dataset contains only abbreviations LW - left wing, ST- stricker, etc.

For 4335 players were added Leagues and Countries where they play, based on the name of club they belong. Those Leagues are in Europe and the most known by average person: Bundesliga (Germany), League 1 (France), Serie A (Italy).

## [1] "character"

2. Vizualisations

After data was cleaned, it’s time to find something interesting.

2.1 Skills analysis

As we have a lot of skills, I want to see they are correlated. For that it’s easy to create a heatmap. Here I will use the whole dataset with 18k players.

I see some interesting correlations. I assume that “interesting” is greater than 0.7 or very red on this graph.

For example: correlation between Acceleration and Agility, Balance, Dribbling, Sprint Speed. Let’s investigate.

Agility is the same in football as it is in any other sport. It’s about how quickly player can change direction without it affecting his/her balance. And it’s is highly correlated with Acceleration,

As it’w written above Agility & Balance are also super tight. And that can be seen on the scatterplot.

Dribbling at lower acceleration is as much connected to acceleration, but when both variables rise, it’s clearly seen that they come together.

And of course Acceleration and Sprint Speed are almost the same numbers.

2.2 Age

Distribution & The Average Age of The Players in each League For those plots I used the smaller dataset only with Leagues.

Average age is quite similar. The youngest player is 15 and the oldest 45.

The Age doesn’t affect very much the Wage.

The Overall rating increase as the Age increases.

Quite obvious that the older player, the slower he becomes.

2.3 Preferred Foot: Does it matter?

## 
##  Left Right 
##  4211 13948

There are way more football players that prefer their right foot.

Left foot has a little bit better result but not significantly.

But wage is the same for all players.

2.4 Map of countries

The most of the players come from Europe.

2.5 Leagues analysis

As Premier League (England), is the most expensive, let’s dive deeper into start of the league.

Looks like there are no young start in Munich. Everyone is mature and has the same overall and potential.

Around 25 all players reach their potentials. Let’s look in those who havent’ reached their potential yet.

A lot of contracts will expire soon. Probably in 2020 there will be a lot of interesting transfers.

## [1] 4

Even there are players who are in one team for 19 years, the average sitribution stays quite the same.

There is no huge correlation between position and reputation. Those outliers are Ronaldo and Messi, Gigi Buffon as Goalkeeper. Everyone else follows quite the same pattern.

Conclusion.

There are a lot of insights in the data. I haven’t checked height and weight, BMI and body type. Probably there is also a lot to discover.

I believe the most useful is Shiny App with Scouting possibility. And also searching the best players, to form “dream team”