Notational Analysisa Math Perspective

Notational analysis a mathematical perspective
Analysis Modelling Feedback
Observation Performance
Improvement
Mike Hughes, Benn Blackburn and Nic James
Research Data Performance Analyst
Coach/Athlete
Linear relationships in data gathering and feedback

7th Australasian Conference on Mathematics and Computers in Sport
Performance Analyst
Biomechanist Gathering systems DATA Processing systems Biomechanist
Notational Analyst Gathering systems DATA Processing systems Notational Analyst
Motor Control Gathering systems DATA Processing systems Motor Control
Coach/Athlete
The role of the performance analyst using early analogue video and computer systems
Performance Analysis Team
Motor Control Coach Athletes
Coach Athletes
Performance Data
Notational Analyst
Biomechanis t
Coach Athletes
A digital systems approach to the data sharing that the interactive commercial systems have enabled for performance analysts working with coaches and athletes (apologies to Popper).
ANALYST
The answer to the mystery of the universe?

ANALYST PERFORMANCE INDICATORS
Performance Indicators
Introduction
What are:PERFORMANCE INDICATORS? (Hughes and Bartlett, 2002) Performance Indicators are a selection or combination of action variable(s) that aim to define some aspect, or all, of a performance.
Introduction
Why use:PERFORMANCE INDICATORS? Performance Indicators are used to assess performance either comparatively, with previous performances, or absolutely.
Notational Analysis - Performance Indicators
Examples:SOCCER : Shots, Passes, Passing Accuracy RUGBY : Turnovers, Tackles, Passes/Possession BADMINTON : W/E ratio, shots/rally, Quality serve/return CRICKET : Strike rate, Dismissal rate, Fielding Efficiency
Performance Indicators?
How do we choose the performance indicators?

Notational Analysis - Performance Indicators Definition of Success:It could be defined by winning (scoring more goals than the opposition) but it may not. Or a coach may be looking for a qualitative improvement in performance - which could be assessed by a performance indicator.
Notational Analysis - Performance Indicators Definition of Success:Success then is relative either to your opposition, racket sports more winners; less errors, invasive games more points or goals than the opposition or to previous performances of your own team, or individual player, or to aggregated means of peer performances
Types of Performance Indicators:-
SCORING : Goals etc., W, E, W/E, Goals/Shots, Dismissal rate, etc. QUALITY : Turnovers, Tackles, Passes/Possession, shots/rally, Strike rate, etc. POSITIVE or NEGATIVE
Notational Analysis - Performance Indicators Dangers of Performance Indicators:-
In other areas of science, performance indicators tend to be ratios of variables, or combinations of variables, that then render the final P.I. dimensionless:-
E.g. In aerodynamics, Mach No. = Velocity of aircraft Velocity of sound In fluid dynamics, Reynolds No. = Ud =density U=velocity =viscosity d=size of object
Performance Indicators - GENERIC

Match Classification Technical Tactical
Data for both teams
(TV)N / (TV)TOTAL or
Means of peer performances
(TV)N / (POSSn) TOTAL
Dangers of Performance Indicators:Performance Indicators should be normalised or standardised in some way to the respective performance variables, and should also be used comparatively with either your opponents data, previous data of your own performances, or with aggregated data of performances of your own level.
Notational Analysis - Performance Indicators To differentiate between important data and DIRTY WASHING:*Choose parameters that relate strongly to *Are there ways of outcome or quality of combining variables in performance. a group that will say more about this performance? *Can these be related to other variables or previously calculated means? *What are the units of your Performance Indicator?
Dangers of Performance Indicators:-
A single action variable, taken in isolation, can give distorted impression of a performance because of other variables, more or less important. E.g. TEAM A: 12 Turnovers; TEAM B: 8 Turnovers TEAM B playing better than TEAM A?
Dangers of Performance Indicators:E.g. TEAM A: 12 Turnovers; TEAM B: 8 Turnovers TEAM B playing better than TEAM A? This will depend upon the possession of both the teams - if TEAM A have had twice as many possessions (48) as TEAM B (24) then their relative performance w.r.t. TURNOVERS/POSSESSION (T/P) will be better than that of TEAM B (T/P)A = 1/4; (T/P)B = 1/3
Dangers of Performance Indicators:Let us consider a more complex example
One of the most quoted research studies in notation is that of Reep and Benjamin (1968).
Most frightening is the effect that this study has had on British soccer and its coaching.
They found:80% of goals resulted from a sequence of three passes or less, 60% of all goals came from possession gained in the final attacking third of the pitch, and a goal is scored every 10 shots (approximately).
Hughes, C. (1985,1990) reinforced these ideas and, as he was the Director of coaching for the English Football Association (F.A.), his influence in the game was considerable. So much so, that these tenets were included in the coaching literature produced by the F.A.
Notational Analysis - Performance Indicators Dangers of Performance Indicators:35 30 25 Goals 20 15 10 5 0 0 1 2 3 4 5 6 7 8 > Touches/Possession 1990 1994
Patterns of goal scoring with respect to the different lengths of possessions in the 1990 and 1994 world cups for soccer.
Notational Analysis - Performance Indicators Dangers of Performance Indicators:9000 8000 7000 6000 5000 4000 3000 2000 1000 0 0 1 2 3 4 5 6 7 8 > Touches/Possession
Frequency
1990 1994
Frequency of each possession string in the two tournaments.
Notational Analysis - Performance Indicators Dangers of Performance Indicators:14 12 10

(G/P)*1000
8 6 4 2 0 0 1 2 3 4 5 6 7 8 > Touches/Possession
1990 1994 Mean
Analysis of the number of goals scored per 1000 possessions for the 2 world cups.
Notational Analysis - Performance Indicators Dangers of Performance Indicators:140 120 100

(S/P)*1000
80 60 40 20 0 0 1 2 3 4 5 6 7 8 > Touches/Possession
1990 1994 Mean
Frequency of shots per 1000 possessions for the 1990 and 1994 World Cups.
They found:80% of goals resulted from a sequence of three passes or less, 60% of all goals came from possession gained in the final attacking third of the pitch, and a goal is scored every 10 shots (approximately).
NOT SO SIMPLE
ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?
Multivariate Statistics
Multiple Linear Regression Discriminant Function Analysis Binary Logistic Regression
Multiple Linear Regression

Linear Regression produces an equation in the form
Y = a + b.X
Multiple Linear Regression produces an equation in the form

Y = b0 + b1.X1 + b2.X2 + b3.X3 + + bn.Xn
By using databases from tournaments, European Championships, World Championships, etc., we can assess the relative importance of the PIs selected. That is we are predicting Y (a known outcome) using several PIs - X1 to Xn
Assumptions of MLR (Ntoumanis, 2001)

Ratio of cases to independent (X) variables should be at least 5 : 1 and ideally 20 : 1 All outliers should be excluded or transformed
Save standardised residuals and explore these
Residuals
Should be normally distributed There should be no relationship between any independent (X) variables and residuals There should be no relationship between residuals and predicted values which can be saved Any relationship between the residuals and dependent (Y) variable should be linear The residuals should be independent, ie no relationship between order of observation (time-order) and residual
RELIABILITY
PERFORMANCE
Reliability
A guide to some ideas about some of the issues and problems associated with reliability.
Analysis procedures for non-parametric data from performance analysis MIKE HUGHES, STEVE-MARK COOPER AND ALAN NEVILL (2001, eIJPAS, 2)
RELIABILITY
PERFORMANCE
RELIABILITY
PERFORMANCE
HOW MUCH DATA?
PERFORMANCE PROFILE
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PERFORMANCE PROFILE
Normative Profiles
Some examples of sample sizes for profiling in sport.
Research
Reep & Benjamin (1969) Eniseler et al., (2000) Larsen et al., (2000) Hughes et al., (1988) Tyryaky et al., (2000) Hughes (1986) Hughes & Knight (1993) Hughes & Williams (1987) Smyth et al., (2001) Blomqvist et al., (1998) O'Donoghue (2001) Hughes & Clarke (1995) O'Donoghue & Ingram (2001)
Sport
Soccer Soccer Soccer Soccer Soccer Squash Squash Rugby Union Rugby Union Badminton Badminton Tennis Tennis
N
(matches for profile)
3,216 4 4 8 (16 teams) 4 and 3 (2 groups) 12, 9 & 6 3 groups 400 rallies 5 5 and 5 5 16, 17, 17, 16, 15 400 rallies 1328<rallies<4300 (8 groups)
Performance Profiles
Performance profiling of an elite male badminton player (Hughes, Evans and Wells, 2001)
Search for a normative profile
16 14
12 Cumulative mean of data value
10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Number of a data set Series1
Performance profiling of an elite male badminton player The cumulative means of each variable were examined over a series of matches/games. At the first point, the number of matches, N(E), where the cumulative mean consistently lay within set limits of error was recorded as the establishment of a normative template of play. These limits of error are a percentage deviation (+/- 1%; +/- 5%; +/- 10%) of the overall data mean about the overall mean. Let n g N(E) N(T) = = = = the variable number of matches the variable number of games value of n to reach limits of error total number of matches
Cumulative mean = (Sum of the frequencies of n) / n Limits of error (10%) = Mean N(T) (Mean N(T) x 0.1) Limits of error (5%) = Mean N(T) (Mean N(T) x 0.05) Limits of error (1%) = Mean N(T) (Mean N(T) x 0.01) and Computers in Sport
7th Australasian Conference on Mathematics
Study 2:- Performance profiling of an elite male badminton player
M e a n n u m b e r o f s h o t s p e r r a l l y b y m a t c h 11.0
10 .0
9 .0
8 .0
7 .0
6 .0
5 .0
4 .0 1 2 3 4 5 6 7 8 9 1 11
N u m b e r o f m a t c h e s
C u m u la t iv e m e a n L e s s 10 % le s s 5 %
P lu s 10 % P lu s 5 %
Study 2:- Performance profiling of an elite male badminton player
Fi g u r e 4 . 5 Me a n n u mb e r o f s h o t s b y g a me 500.0
450.0
400.0
350.0
300.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Number of games
Cumulative mean
'+ 10%
'-10%
'+ 5%
'-5%
'+ 1%
'-1%
European
525.00 500.00 475.00 450.00 425.00 400.00 375.00 350.00 325.00 300.00 275.00 250.00 225.00 200.00 175.00 150.00 125.00 100.00 75.00 50.00 25.00 0.00 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Pass Runs Dribbles Crosses Header Shots
Mean score
Number of matches
Number of matches needed to7th Australasian Conference on Mathematics for attacking variables for achieve a normative profile and Computers in Sport European teams
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PERFORMANCE PROFILE
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE
James et al. (2004, JSS, in press)
James et al. (2004) suggested an alternative approach whereby the specific estimates of population means are calculated from the sample data through confidence limits (CLs). CLs represent upper and lower values between which the true (population) mean is likely to fall based on the observed values collected. Calculated CLs naturally change as more data is collected, typically resulting in the confidence interval (CI - upper CL minus lower CL) decreasing. Confidence intervals (CIs) were therefore suggested to be more appropriate as performance guides compared to using mean values. Using a fixed value appears to be too constrained due to potential confounding variables that typically affect performance, making prescriptive targets untenable.
James et al. (2004, JSS, in press)

From a theoretical perspective, James et al. argued that the use of CIs can also add significance to the judgement of the predictive potential of a data set, i.e. whether enough data has been collected to allow a reasonable estimation. For their investigation a criterion was formulated to test the rate of change of the CI for stability. Initially 95% CIs were calculated for each performance indicator as soon as enough match data had been collected (N = 2) and each time more data was added the new CI was calculated. This meant that CIs could be constructed for each performance indicator after 2, 3 and. N matches respectively.
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE COMPARING PROFILES
Comparing profiles
Nevill, A., Atkinson, G., Hughes, M. and Cooper, S-M. (2002). Journal of Sports Science, 20, 829 - 844. Many research papers have used parametric tests in the past these have been found to be slightly less sensitive than the non-parametric tests, and they did not respond to large differences within the data. The results of performance analysis are very often recorded as discrete events. Clearly, investigating categorical differences in discrete data using traditional parametric tests of significance (e.g. ANOVA, based on the continuous symmetric normal distribution) is inappropriate. More appropriate statistical methods are promoted based on two key discrete probability distributions, the Poisson and binomial distributions. 7th Australasian Conference on Mathematics
and Computers in Sport
Comparing profiles
When carrying out tests of significance on continuous data using regression and analysis of variance (ANOVA), the observed random variation is assumed to have a normal distribution. Clearly, the frequency distribution of discrete events, such as the number of shots per rally in tennis or squash, do not follow a normal distribution. For example, the frequency distribution of shots per rally of an elite squash player over a three-game match (total number of rallies = 104) is discrete, positively skewed and not normally distributed. Two key distributions for such discrete data are the Poisson and binomial distributions.
Comparing profiles
Comparing profiles
Two approaches are proposed and compared using examples from notational analysis:The first approach is based on the classic chi-square test of significance (both the goodness-of-fit test and the test of independence). The second approach adopts a more contemporary method based on log-linear and logit models fitted using the statistical software GLIM.
Comparing profiles
Provided relatively simple one-way and two-way comparisons in categorical data are required, both of these approaches result in very similar conclusions. However, as soon as more complex models or higher-order comparisons are required, the approach based on log-linear and logit models is shown to be more effective. Indeed, when investigating those factors and categorical differences associated with binomial or binary response variables, such as the proportion of winners when attempting decisive shots in squash or the proportion of goals scored from all shots in association football, logit models become the only realistic method available. Conference on Mathematics 7th Australasian
and Computers in Sport
Further ?
We need tests of difference that are far more sensitive. The winner of the womens 400m Olympic Gold in Sydney performed 1 - 2% better than the person who was 8th - one is a millionaire - the other?
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE COMPARING PROFILES
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE COMPARING PROFILES MODELLING PERFORMANCE
Modelling
Empirical Modelling Stochastic Modelling Perturbations Artificial Intelligence Expert Systems Neural Networks
Empirical Modelling
Modelling - Probability
Sport and Chance Reep and Benjamin (1968) Ladany and Machol (1977) Alexander et al (1988) McGarry and Franks (1996)
Stochastic Modelling
Problems:too few data antecedent shot is a naive predictor of the next shot to be selected memory-limiting nature of stochastic (Markov) processes, where the future is predicted only from the present, might be an insufficient descriptor of sports behaviours
Problems:Sports analysts have tended to record all the data from a sports contest and to search those data for patterns. Implicit in this method of analysis are two assumptions. The first assumption is that if the data are to have information value then they are likely to be repeated under similar future circumstances. The second assumption is that the data are of equal importance, at least in the long 7th Australasian Conference on Mathematics and Computers in Sport run.
Perturbations
World Congress of Notational Analysis of Sport, Burton Manor, 1992 Downey talked of rhythms in badminton rackets, cooperation, until there was a dislocation of the rhythm a perturbation sometimes resulting in a rally end situation ( a critical incident), sometimes not.
Perturbations
Harmonic Motion
Fig. 3.1. A schematic example of Simple Harmonic Motion (SHM) from Weinstein (2004).
Perturbations
Server 14 Receiver
Distance (m) Rally 21

6 5 4 3 2 1 0 00:14:13:01 00:14:21:12 00:14:29:23 00:14:38:09 00:14:46:20 00:14:55:06 00:15:03:17 00:15:12:03 00:15:20:14 00:15:29:00 00:15:37:11 00:15:45:22 00:15:54:08 00:16:02:19 00:16:11:05
12
10
Distance (m)
Series1
Tim e (s)
0 0.0 1.0 2.0 4.0 5.2 7.2 8.4 T ime ( s) 9.6 11.0 13.0 14.2 15.4 17.0
Figure 1.2. Distance Time Graph for Pilot study 1

Perturbations
If we study a system only in the linear range of its operation where change is smooth, its difficult if not impossible to determine which variables are essential and which are not. Most scientists know about nonlinearity and usually try to avoid it. Here we exploit qualitative change, a nonlinear instability, to identify collective variables, the implication being that because these variables change abruptly, it is likely that they are also the key variables when the system operates in the linear range. Scott Kelso, 1999
Perturbations
SQUASH McGarry, Khan & Franks, 1999
SOCCER Hughes, Dawkins, Davids & Mills, 1998

Artificial Intelligence
- Roger Bartlett
.and Jurgen Perl!
The Decision Making Scope of AI

Expert systems : Rule-based. Fuzzy. Frame-based. Artificial neural networks : Biological and artificial neural networks. The Perceptron. Multi-layer neural networks. Recurrent neural networks. Self-organising neural networks.
Expert Systems
Advantages
Separate knowledge from processing, unlike conventional programs. Provide an explanation facility. Can deal with incomplete and vague data. Can model fuzzy human decision-making. Are good for diagnosis. Shells for development of expert systems are widely available (e.g. add-ons to MATLAB).
Disadvantages
Need to acquire knowledge from experts; this is a major problem. Very domain-specific; fast bowling one could not be used for javelin throwing. Opaque relationships between rules. In general, do not have an ability to learn. Have to manage conflicts between rules. Ineffective rules searching Conference onthrough all rules in each trawl Mathematics 7th Australasian and Computers in Sport cycle.
Expert Systems - Performance Analysis

Given that they are good diagnostic tools and that system shells easily available, how widespread is the use of them in PA? Not very! The reality conflicts with the positive view of their utility by Lapham and Bartlett in 1995(3).
Artificial Neural Networks

Allow computers to learn from experience and by anology(1). A computer program that tries to create a mathematical model of neurons in the brain(2). An interconnection of simple adaptable processing elements or nodes(2): Nodes simplified models of brain neurons. Store experiential knowledge as pattern of connected nodes and synaptic weightings between them. Non-linear programs that represent non-linear systems, such as the human movement system and games. Originally developed to exploit the power of parallel 7th Australasian based. processing, now mostly PCConference on Mathematics and Computers in Sport
Advantages
Learn by experience; in the case of self-organising ANNs, without a teacher! Are good for classification, clustering and prediction tasks. Can be adapted for inexact or incomplete data through fuzzy ANNs. Are widely available, e.g. the MATLAB Neural Network Toolbox, and relatively simple programs. Seem to mimic brain processes. Provide link to dynamic systems theory as non-linear program representations of non-linear biological systems.
Disadvantages
They are opaque black boxes with no explanation of the reasoning process. The rules within the non-linear network are not well understood; the non-linear characteristics may prohibit simple and understandable rules(1). To validate their output, they need test cases for which output is known. They often do not work well for inputs outside the range used for learning. Back propagation is very slow, although widely used for pattern recognition. Kohonen SOMs need lots of learning data and aren't dynamic.
ANNS in Performance Analysis

As their main function is for classification, clustering and prediction, and that they are now easily available (but only recently) - how widespread is their use in PA? NOT VERY They have been used in PA, both in technique analysis and notational analysis, and in other branches of sport and exercise science.
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE COMPARING PROFILES MODELLING PERFORMANCE
RELIABILITY
PERFORMANCE
HOW MUCH DATA? EMPIRICAL METHODS PREDICTION FROM VARIANCE PERFORMANCE PROFILE COMPARING PROFILES MODELLING PERFORMANCE PREDICTION
Performance Prediction
Multiple Linear Regression Discriminant Function Analysis Binary Logistic Regression Neural Network
Example 2003 Rugby World Cup

Peter O'Donoghue and Jason Williams
Individual Human Predictions Expert focus group MLR BLR ANN Simulation package

Points difference between higher and lower ranked teams is the Y variable The X variables are Difference in ranking points Difference in distance travelled to tournament Difference in recovery days since last match Used 137 cases from 1987 to 1999 to do the linear regression model in SPSS Used 40 group matches to see predictions in Excel

Rugby union is an easier sport than soccer to predict the outcome for because there is a greater amount of scoring in rugby union and currently less strength in depth in the international game. The most successful machine based method was the simulation package which produced a prediction that recognised the effect of combined conditional probabilities on the overall outcome of the tournament. Quantitative and computer-based prediction methods were more successful at predicting the results of the 2003 Rugby World Cup than most of the predictions made by individual humans which were based on qualitative analysis. However, the expert focus group demonstrated that 7th Australasian Conference on Mathematics human expertise still exceedsinthat of machine based and Computers Sport methods.
RELIABILITY
EMPIRICAL METHODS
PREDICTION FROM VARIANCE
Performance Indicators? Reliability Discriminate between PIs Empirical Profiling Predictive Profiling Methods? Comparing data Modelling? Prediction?
PERFORMANCE WHICH ARE MOST IMPORTANT? HOW MUCH DATA? PERFORMANCE PROFILE COMPARING PROFILES MODELLING PERFORMANCE PREDICTION
ANALYST
The answer to the mystery of the universe?

ANALYST
To get the correct answers you 42? have to ask the right questions.
International Society of Performance Analysis of Sport Mike Hughes International Society of Computers in Sport Science Jurgen Perl E-JournalsInternational Journal of Performance Analysis of Sport International Journal of Computers in Sport Science mhughes@uwic.ac.uk
Notational analysis a mathematical perspective.
THANKS for your attention.
Mike Hughes, Centre for Performance Analysis, University of Wales Institute Cardiff.
Papers
Hughes, M.D. (2004). Performance Analysis a mathematical perspective. EIJPAS, International Journal of Performance Analysis Sport (Electronic), 4, 2, 97 139. Hughes, M.D. and Bartlett, R.(2002). The use of performance indicators in performance analysis. Journal of Sports Science 20, 739 754.

Notational Analysisa Math Perspective

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Notational Analysisa Math Perspective

Uploaded by

Copyright:

Available Formats

Notational analysis a mathematical perspective

Analysis Modelling Feedback

Mike Hughes, Benn Blackburn and Nic James

Research Data Performance Analyst

Linear relationships in data gathering and feedback

Biomechanist Gathering systems DATA Processing systems Biomechanist

Notational Analyst Gathering systems DATA Processing systems Notational Analyst

Motor Control Gathering systems DATA Processing systems Motor Control

Performance Analysis Team

Motor Control Coach Athletes

The answer to the mystery of the universe?

ANALYST PERFORMANCE INDICATORS

7th Australasian Conference on Mathematics and Computers in Sport

7th Australasian Conference on Mathematics and Computers in Sport

Notational Analysis - Performance Indicators

How do we choose the performance indicators?

Notational Analysis - Performance Indicators

Types of Performance Indicators:-

Notational Analysis - Performance Indicators Dangers of Performance Indicators:-

Performance Indicators - GENERIC

Data for both teams

Means of peer performances

(TV)N / (POSSn) TOTAL

7th Australasian Conference on Mathematics and Computers in Sport

Notational Analysis - Performance Indicators

Notational Analysis - Performance Indicators

Dangers of Performance Indicators:-

Notational Analysis - Performance Indicators

Notational Analysis - Performance Indicators

Dangers of Performance Indicators:Let us consider a more complex example

Notational Analysis - Performance Indicators Dangers of Performance Indicators:-

Frequency of each possession string in the two tournaments.

Notational Analysis - Performance Indicators Dangers of Performance Indicators:14 12 10

1990 1994 Mean

Notational Analysis - Performance Indicators Dangers of Performance Indicators:140 120 100

1990 1994 Mean

Notational Analysis - Performance Indicators Dangers of Performance Indicators:-

ANALYST PERFORMANCE INDICATORS

7th Australasian Conference on Mathematics and Computers in Sport

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

7th Australasian Conference on Mathematics and Computers in Sport

Multiple Linear Regression Discriminant Function Analysis Binary Logistic Regression

Multiple Linear Regression

Multiple Linear Regression produces an equation in the form

Assumptions of MLR (Ntoumanis, 2001)

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

7th Australasian Conference on Mathematics and Computers in Sport

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

7th Australasian Conference on Mathematics and Computers in Sport

7th Australasian Conference on Mathematics and Computers in Sport

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

7th Australasian Conference on Mathematics and Computers in Sport

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

HOW MUCH DATA?

7th Australasian Conference on Mathematics and Computers in Sport

ANALYST PERFORMANCE INDICATORS WHICH ARE MOST IMPORTANT?

HOW MUCH DATA? EMPIRICAL METHODS PERFORMANCE PROFILE

7th Australasian Conference on Mathematics and Computers in Sport

7th Australasian Conference on Mathematics and Computers in Sport

12 Cumulative mean of data value

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Number of a data set Series1

7th Australasian Conference on Mathematics and Computers in Sport

7th Australasian Conference on Mathematics and Computers in Sport