StandardStandard

The Identification of ‘Game Changers’ in England Cricket’s Developmental Pathway for 3 Elite Spin Bowling: A Machine Learning Approach. / Jones, Benjamin; Hardy, Lewis; Lawrence, Gavin et al.
Yn: Journal of Expertise, Cyfrol 2, Rhif 2, 06.2019, t. 92-120.

Allbwn ymchwil: Cyfraniad at gyfnodolynErthygladolygiad gan gymheiriaid

HarvardHarvard

APA

CBE

MLA

VancouverVancouver

Author

RIS

TY - JOUR

T1 - The Identification of ‘Game Changers’ in England Cricket’s Developmental Pathway for 3 Elite Spin Bowling: A Machine Learning Approach

AU - Jones, Benjamin

AU - Hardy, Lewis

AU - Lawrence, Gavin

AU - Kuncheva, Ludmila

AU - Brandon, Raphael

AU - Such, Peter

AU - Bobat, Mo

PY - 2019/6

Y1 - 2019/6

N2 - Research exploring the development of expertise has mostly adopted linear methods to identify precursors of expertise, assessing statistical differences between groups of isolated variables by way of attaching importance to variables, e.g., deliberate practice hours (Ericsson et al., 1993). However, confining the complex nature of expertise development to linear investigations alone may be overly simplistic. Consequently, to better understand the multidimensional and complex nature of expertise development, we applied (non-linear) pattern recognition analyses to a set of 93 features obtained from a sample of 15 elite (International) and 13 sub-elite (First-class county) cricket spin bowlers. Our study revealed that a subset of 12 developmental features, from a possible 93, discriminated between the elite and sub-elite groups, with very good accuracy. The 12-feature subset forms a holistic development profile, reflecting the elite’s earlier engagement in cricket, greater quantity of domain-specific practice and competition, and superior adaptability to new levels of senior competition. Evidence for the external validity of this new model is offered by its ability to correctly classify data obtained from five unseen spin bowlers with 100% accuracy. After consideration of these quantitative findings, the content of qualitative data provided by the cricketers was subsequently analysed to obtain a deeper understanding of the features that discriminate between the elite and sub-elite groups.

AB - Research exploring the development of expertise has mostly adopted linear methods to identify precursors of expertise, assessing statistical differences between groups of isolated variables by way of attaching importance to variables, e.g., deliberate practice hours (Ericsson et al., 1993). However, confining the complex nature of expertise development to linear investigations alone may be overly simplistic. Consequently, to better understand the multidimensional and complex nature of expertise development, we applied (non-linear) pattern recognition analyses to a set of 93 features obtained from a sample of 15 elite (International) and 13 sub-elite (First-class county) cricket spin bowlers. Our study revealed that a subset of 12 developmental features, from a possible 93, discriminated between the elite and sub-elite groups, with very good accuracy. The 12-feature subset forms a holistic development profile, reflecting the elite’s earlier engagement in cricket, greater quantity of domain-specific practice and competition, and superior adaptability to new levels of senior competition. Evidence for the external validity of this new model is offered by its ability to correctly classify data obtained from five unseen spin bowlers with 100% accuracy. After consideration of these quantitative findings, the content of qualitative data provided by the cricketers was subsequently analysed to obtain a deeper understanding of the features that discriminate between the elite and sub-elite groups.

M3 - Article

VL - 2

SP - 92

EP - 120

JO - Journal of Expertise

JF - Journal of Expertise

SN - 2573-2773

IS - 2

ER -