Alberto Trindade Tavares - University of Wisconsin–Madison
Download
Report
Transcript Alberto Trindade Tavares - University of Wisconsin–Madison
ECE/CS/ME 539 - Introduction to Artificial Neural Network and Fuzzy Systems
Alberto Trindade Tavares
Brazilian Soccer League
Since 2003, the Brazilian Soccer League has the following format:
20 participating clubs
Each club faces every other club twice in the season, once at their home
stadium, and once at that of their opponents
380 matches divided into two parts:
First half: May-August
Second half: September-December
A match has three possible results:
Win of the home team
Draw
Loss of the home team
Goal of this Work
Predict the outcome (win of home team, draw, or loss of home team) of
every game of the second half for the current season (2013)
Using as training data the game results of the first half of 2013 season
Develop two classifiers, using MATLAB, for performing these
predictions :
Maximum Likelihood Classifier
Multi-Layer Perceptron
Compare their accuracy between themselves and to other works
Feature Vector
The feature vector for representing a match instance has six features,
the first three for the first team (home), and last three for the second
team (visiting):
# wins as
home team
# draws as
home team
First Team
# losses as
home team
# wins as
# draws as
# losses as
visiting team visiting team visiting team
Second Team
Results from 2003 season to the last match of current season
Data Extraction
Extraction of results of every match since 2003
Two different sources:
2003-2004 seasons: http://www.bolanaarea.com/gal_brasileirao.htm
2005-2013 seasons: http://www.campeoesdofutebol.com.br
Python program for parsing the HTML pages, and storing the results
into text files, which can be read via MATLAB function load
Maximum Likelihood Classifier
Gaussian Distribution
P(x)
Win
Loss
Draw
x
Maximum Likelihood Classifier (Results)
Classification rate per round:
Average Classification Rate = 53.1579%
Maximum Likelihood Classifier (Results)
Total Confusion Matrix:
Predicted Wins
Predicted Draws
Predicted Losses
Actual Wins
71
8
18
Actual Draws
25
9
15
Actual Losses
19
4
21
Multi-Layer Perceptron
# Hidden Layers = 3
# Neurons in First Hidden Layer = 3
# Neurons in First Hidden Layer = 20
# Neurons in First Hidden Layer = 3
Learning rate (α) = 0.1
Momentum = 0
Hidden layers use hyperbolic tangent activation function, and output
layer uses sigmoid activation function
Multi-Layer Perceptron(Results)
Classification rate per round (10 runs):
Average Classification Rate = 55.7895%
Multi-Layer Perceptron(Results)
Total Confusion Matrix:
Predicted Wins
Predicted Draws
Predicted Losses
Actual Wins
78
7
12
Actual Draws
27
11
11
Actual Losses
23
4
17
Comparison with other work
A. Joseph, N.E. Fenton, M. Neil. Predicting football results using
Bayesian nets and other machine learning techniques (2006)
Published in the Journal Knowledge-Based Systems
Their results:
Naïve BN: 47.86%
kNN: 50.58%
Expert BN: 59.21%
My Results:
Maximum Likelihood: 53.1579%
Multi-Layer Perceptron: 55.7895%
Questions?