| KEYWORD |
Development of an algorithm for the fantasy football game able to forecast the football players performance through machine learning techniques
keywords EVOLUTIONARY ALGORITHMS, FANTASY FOOTBALL, MACHINE LEARNING
Reference persons EDGAR ERNESTO SANCHEZ SANCHEZ
External reference persons Niccolo' Battezzati
Teamies Co-founder
www.teamies.it
Research Groups GR-05 - ELECTRONIC CAD & RELIABILITY GROUP - CAD
Description the fantasy football game is one of the most played games both in Italy and also worldwide. Currently, it is possible to find several on-line platforms where to play and win prizes. One of the most difficult tasks to be performed in this game is to select the players to engage and line-up at every match; in fact, there are different forums, on-line sites, and social pages where the fantasy football coaches can ask for advise.
The main goal of this thesis is to design and implement a machine learning algorithm able to forecast the footballers performance in order to advise the fantasy football players the most suitable team to line-up at every match, exploiting the players statistics available on-line.
The overall thesis work is to be divided in two steps: the first one is oriented to analyze and tune up an evolutionary algorithm able to optimize the configuration parameters of a heuristic algorithm developed by a fantasy football expert. The second step is mainly oriented to develop a neural network based tool able to forecast footballers performances without any statistical information about the players.
Required skills the thesis development requires very good programming skills in a high-level language (i.e., C++, Python), and basic knowledge on HTML and SQL. Preliminary knowledge on the fantasy game as well as machine learning techniques are not mandatory. However, it is advisable to count with passion for sports and new technologies.
Notes the thesis is developed in cooperation with the Teamies startup, incubated at I3P, which provides its experience on the game and design of modern software platforms.
Deadline 05/03/2019
PROPONI LA TUA CANDIDATURA