Feature Extraction and Aggregation for Predicting the Euro 2016 - - PowerPoint PPT Presentation

▶
feature extraction and aggregation for predicting the
SMART_READER_LITE
LIVE PREVIEW

Feature Extraction and Aggregation for Predicting the Euro 2016 - - PowerPoint PPT Presentation

Jun 21, 2023 39 likes •239 views

Feature Extraction and Aggregation for Predicting the Euro 2016 Maryam Tavakol Hamid Zafartavanaelmi, and Ulf Brefeld Riva del Garda, Sep 19, 2016 Agenda Introduction Feature Extraction Prediction & Learning Performance

slide-1
SLIDE 1

Feature Extraction and Aggregation for Predicting the Euro 2016

Maryam Tavakol Hamid Zafartavanaelmi, and Ulf Brefeld

Riva del Garda, Sep 19, 2016

slide-2
SLIDE 2

Agenda

  • Introduction
  • Feature Extraction
  • Prediction & Learning
  • Performance Analysis
  • Summary

2

slide-3
SLIDE 3

Introduction

3

slide-4
SLIDE 4

Feature Extraction

  • Based on available data from the past tournaments
  • General country data
  • FIFA ranking, FIFA points, UEFA ranking, etc.
  • Normalising features using min and max rescaling —keep

the order

4

slide-5
SLIDE 5

Feature Extraction

  • Player specific data
  • Market value, age, num of matches/goals, etc.
  • Obtaining the current squads
  • Goal/play ratio —host advantage for France
  • Averaging for all players of a team
  • Normalising features using min and max rescaling

5

slide-6
SLIDE 6

Add a New Feature

6

slide-7
SLIDE 7

Club Division

Juventus Club rank = 2 Lazio

Club rank = 212

…

7

slide-8
SLIDE 8

Team-Club Harmony

(Normalised Club rank) x (num of players)

8

Country Num of Players Club Club Rank Spain 5 Barcelona 1 Italy 6 Juventus 2 France 2 Juventus 2 Germany 5 Bayern Munich 4 Belgium 3 Liverpool 42 Poland 3 Legia 52 Portugal 4 Sporting CP 179 Wales 3 Crystal Palace 0* Iceland 2 Hammarby 0*

slide-9
SLIDE 9

Prediction

  • A score per country is defined as a weighted sum
  • f features, i.e., linear function
  • The probabilities are computed based on obtained

scores

9

si = θ>

i xi

slide-10
SLIDE 10

Prediction

10

Win probability for team i Lose probability for team j Probability of draw

slide-11
SLIDE 11

Learning

  • Capture the outcome probabilities from the head to

head record of pair of countries

  • Germany vs. France: 27 times
  • 10 win for Germany, 12 for France and 5 draw

11

pwG = 10 27, pwF = 12 27, pd = 5 27

slide-12
SLIDE 12

Learning

  • Converting probabilities to scores
  • Obtaining parameters from the closed form solution
  • f ridge regression problem

12

ˆ θ = (X>X + I)1X>ˆ s

slide-13
SLIDE 13

Performance Analysis

  • Compare prediction results to actual tournament
  • utcome
  • Until Quarter-Final (QF)
  • Evaluation by multi class logarithmic loss

13

Logloss = − 1 N

N

X

i=1 M

X

j=1

yij ∗ log(pij)

slide-14
SLIDE 14

Overal Performance

  • Error of prediction for 45 matches before QF
  • Average error: 1.3187

14

log loss

slide-15
SLIDE 15

Insufficient Data

  • Relation of performance with amount of historical

data

15

Num of historical data Error per country

log loss num

slide-16
SLIDE 16

Sufficient Data

  • Reduction of error from 1.3187 to 1.1129 for teams

with more than 4 historical records

16

log loss

slide-17
SLIDE 17

Role of Past Euros

  • Eliminating teams with less than 2 appearance in

past Euro cups, error: 0.9680

17

log loss

slide-18
SLIDE 18

Baseline

  • Compare to a simple baseline (based on FIFA

ranking only)

18

log loss

slide-19
SLIDE 19

Summary

  • Collecting data
  • Feature extracting/cleaning
  • New feature: team-club harmony
  • Learn a linear model
  • Effect of historical data on the performance

19

slide-20
SLIDE 20

Questions?

Thanks for your attention

Email: tavakol@leuphana.de