Football is one of the most popular games in the world. A whole line of business has been built from broadcasting rights to the sale of players. As a team can be seen as a business venture from the point of view of owners and supporters, it is no coincidence that they are trying to support and improve the performance of the teams with state-of-the-art methods. As a result, information technology and data-based performance measurement and decision support in strategy development play a growing role.
In my thesis, I look at a football database on the internet, which serves as a digital record of a match. After a brief overview, I present the current status of football analysis and examine the techniques and the available soccer analysis softwares. Form the given data set with different preprocess and data transforming techniques I transformed it for a predictable data set. Besides these I made a possible visualization of a match with available tools.
After this I have completed a detailed analysis of the available data set, furthermore with help from different machine-learning methods (decision tree, support vector machine, neural network, etc.) have predict the outcome of the match from the first half result (Team A Won, Draw, Team B Won).
The performance of developed algorithms was performed on real data, which proves that the outcome of the match is up to 83.72 percent from the half-time statistics.