Neural network to predict dutch soccer league

A multi lever perceptron to predict the chances (0-100%) of each team
in this weeks eredivision soccer  league

depends on numpy, pandas and keras
To get the data:

data from http://www.football-data.co.uk/netherlandsm.php
this week : http://www.football-data.co.uk/matches.php

(It downloads it now automatically but probably the links need to be changed for the new season(currently 2016-2017)

Output looks like this(numbers are percentage between 0 and 100):


match home draw away
Den Haag-Twente 28.0 27.0 45.0
Excelsior-Willem II 30.0 39.0 32.0
Groningen-Utrecht 25.0 38.0 38.0
Heerenveen-Roda 48.0 33.0 20.0
Nijmegen-Sparta Rotterdam 58.0 26.0 16.0
Ajax-Heracles 83.0 12.0 5.0
AZ Alkmaar-Zwolle 47.0 33.0 21.0
Feyenoord-PSV Eindhoven 77.0 13.0 10.0
Go Ahead Eagles-Vitesse 27.0 33.0 40.0




Any questions?

remko66@gmail.com