Marathon Race Planning: A Case-Based Reasoning Approach

Smyth, Barry; Cunningham, Pádraig

doi:10.24963/ijcai.2018/754

Cited by 14 publications

(3 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although such data was not available in our dataset, the increasingly widespread adoption of mobile devices, smartwatches, and wearable sensors [55,56] has the capacity to generate large volumes of additional data (heart-rate, cadence, and power), which may be useful in this regard in the future [57,58]. Already, the availability of such diverse sources of data is enabling several new types of health and fitness applications [59][60][61][62][63] and the emergence of powerful new machine learning techniques has been used to support a variety of related prediction and planning tasks in several sporting domains [64][65][66][67][68][69][70][71][72][73] It is also worth noting that the model of the wall analysed here is defined by a pair of parameters-degree of slowdown and length of slowdown-with specific values-0.25 and 5km, respectively-and it is reasonable to question whether the results would be different if different values had been chosen. We have considered several alternative sets of values and, within reasonable levels of tolerance, there is no material change to the nature of the results as presented.…”

Section: Limitationsmentioning

confidence: 99%

How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon

Smyth

2021

PLoS ONE

Self Cite

View full text Add to dashboard Cite

Introduction In the marathon, how runners pace and fuel their race can have a major impact on race outcome. The phenomenon known as hitting the wall (HTW) refers to the iconic hazard of the marathon distance, in which runners experience a significant slowing of pace late in the race, typically after the 20-mile mark, and usually because of a depletion of the body’s energy stores. Aim This work investigates the occurrence of significant late-race slowing among recreational marathoners, as a proxy for runners hitting the wall, to better understand the likelihood and nature of such slowdowns, and their effect on race performance. Methods Using pacing data from more than 4 million race records, we develop a pacing-based definition of hitting the wall, by identifying runners who experience a sustained period of slowing during the latter stages of the marathon. We calculate the cost of these slowdowns relative to estimates of the recent personal-best times of runners and compare slowdowns according to runner sex, age, and ability. Results We find male runners more likely to slow significantly (hit the wall) than female runners; 28% of male runners hit the wall compared with 17% of female runners, χ2(1, N = 1, 928, 813) = 27, 693.35, p < 0.01, OR = 1.43. Such slowdowns are more frequent in the 3 years immediately before and after a recent personal-best (PB) time; for example, 36% of all runners hit the wall in the 3 years before a recent PB compared with just 23% in earlier years, χ2(1, N = 509, 444) = 8, 120.74, p < 0.01, OR = 1.31. When runners hit the wall, males slow more than females: a relative slowdown of 0.40 vs. 0.37 is noted, for male and female runners, when comparing their pace when they hit the wall to their earlier race (5km-20km) pace, with t(475, 199) = 60.19, p < 0.01, d = 0.15. And male runners slow over longer distances than female runners: 10.7km vs. 9.6km, respectively, t(475, 199) = 68.44, p < 0.01, d = 0.17. Although, notably the effect size of these differences is small. We also find the finish-time costs of hitting the wall (lost minutes) to increase with ability; r2(7) = 0.91, p < 0.01 r2(7) = 0.81, p < 0.01 for male and female runners, respectively. Conclusions While the findings from this study are consistent with qualitative results from earlier single-race or smaller-scale studies, the new insights into the risk and nature of slowdowns, based on the runner sex, age, and ability, have the potential to help runners and coaches to better understand and calibrate the risk/reward trade-offs that exist as they plan for future races.

show abstract

Section: Limitationsmentioning

confidence: 99%

How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon

Smyth

2021

PLoS ONE

Self Cite

View full text Add to dashboard Cite

show abstract

“…What is less well developed, however, is the translation of a goal-time into a specific race strategy and a concrete set of pacing recommendations. We have recently addressed this dual problem of goal-time prediction and pacing recommendation for marathons, by using case-based reasoning (Smyth and Cunningham 2017;2018b;2018a). In short, the goal-time and pacing plan for a target runner is adapted from the race-times and pacing profiles of runners with similar race histories.…”

Section: Goal-time Prediction and Pacing Planningmentioning

confidence: 99%

Recommender Systems: A Healthy Obsession

Smyth

2019

AAAI

Self Cite

View full text Add to dashboard Cite

We propose endurance sports as a rich and novel domain for recommender systems and machine learning research. As sports like marathon running, triathlons, and mountain biking become more and more popular among recreational athletes, there exists a growing opportunity to develop solutions to a number of interesting prediction, classification, and recommendation challenges, to better support the complex training and competition needs of athletes. Such solutions have the potential to improve the health and well-being of large populations of users, by promoting and optimising exercise as part of a productive and healthy lifestyle.

show abstract

“…More sophisticated reasoning is enabled by considering a snapshot of the current case base in part or as a whole. Examination of this information can determine, for example, whether a case is worth adding to a case base because it increases the competence of the CBR system or whether a solution can be discarded without affecting competence (Smyth and Keane 1995). As another example see Reinartz et al contributions in (Reinartz et al,.…”

Section: Maintenance Data Collectionmentioning

confidence: 99%

A dynamic adaptive framework for improving case-based reasoning system performance

Orduña

View full text Add to dashboard Cite

An optimal performance of a Case-Based Reasoning (CBR) system means, the CBR system must be efficient both in time and in size, and must be optimally competent. The efficiency in time is closely related to an efficient and optimal retrieval process over the Case Base of the CBR system. Efficiency in size means that the Case Library (CL) size should be minimal. Therefore, the efficiency in size is closely related to optimal case learning policies, optimal meta-case learning policies, optimal case forgetting policies, etc. On the other hand, the optimal competence of a CBR system means that the number of problems that the CBR system can satisfactorily solve must be maximum. To improve or optimize all three dimensions in a CBR system at the same time is a difficult challenge because they are interrelated, and it becomes even more difficult when the CBR system is applied to a dynamic or continuous domain (data stream). In this thesis, a Dynamic Adaptive Case Library framework (DACL) is proposed to improve the CBR system performance coping especially with reducing the retrieval time, increasing the CBR system competence, and maintaining and adapting the CL to be efficient in size, especially in continuous domains. DACL learns cases and organizes them into dynamic cluster structures. The DACL is able to adapt itself to a dynamic environment, where new clusters, meta-cases or prototype of cases, and associated indexing structures (discriminant trees, k-d trees, etc.) can be formed, updated, or even removed. DACL offers a possible solution to the management of the large amount of data generated in an unsupervised continuous domain (data stream). In addition, we propose the use of a Multiple Case Library (MCL), which is a static version of a DACL, with the same structure but being defined statically to be used in supervised domains. The thesis work proposes some techniques for improving the indexation and the retrieval task. The most important indexing method is the NIAR k-d tree algorithm, which improves the retrieval time and competence, compared against the baseline approach (a flat CL) and against the well-known techniques based on using standard k-d tree strategies. The proposed Partial Matching Exploration (PME) technique explores a hierarchical case library with a tree indexing-structure aiming at not losing the most similar cases to a query case. This technique allows not only exploring the best matching path, but also several alternative partial matching paths to be explored. The results show an improvement in competence and time of retrieving of similar cases. Through the experimentation tests done, with a set of well-known benchmark supervised databases. The dynamic building of prototypes in DACL has been tested in an unsupervised domain (environmental domain) where the air pollution is evaluated. The core task of building prototypes in a DACL is the implementation of a stochastic method for the learning of new cases and management of prototypes. Finally, the whole dynamic framework, integrating all the main proposed approaches of the research work, has been tested in simulated unsupervised domains with several well-known databases in an incremental way, as data streams are processed in real life. The conclusions outlined that from the experimental results, it can be stated that the dynamic adaptive framework proposed (DACL/MCL), jointly with the contributed indexing strategies and exploration techniques, and with the proposed stochastic case learning policies, and meta-case learning policies, improves the performance of standard CBR systems both in supervised domains (MCL) and in unsupervised continuous domains (DACL). El rendimiento óptimo de un sistema de razonamiento basado en casos (CBR) significa que el sistema CBR debe ser eficiente tanto en tiempo como en tamaño, y debe ser competente de manera óptima. La eficiencia temporal está estrechamente relacionada con que el proceso de recuperación sobre la Base de Casos del sistema CBR sea eficiente y óptimo. La eficiencia en tamaño significa que el tamaño de la Base de Casos (CL) debe ser mínimo. Por lo tanto, la eficiencia en tamaño está estrechamente relacionada con las políticas óptimas de aprendizaje de casos y meta-casos, y las políticas óptimas de olvido de casos, etc. Por otro lado, la competencia óptima de un sistema CBR significa que el número de problemas que el sistema puede resolver de forma satisfactoria debe ser máximo. Mejorar u optimizar las tres dimensiones de un sistema CBR al mismo tiempo es un reto difícil, ya que están relacionadas entre sí, y se vuelve aún más difícil cuando se aplica el sistema de CBR a un dominio dinámico o continuo (flujo de datos). En esta tesis se propone el Dynamic Adaptive Case Library framework (DACL) para mejorar el rendimiento del sistema CBR especialmente con la reducción del tiempo de recuperación, aumentando la competencia del sistema CBR, manteniendo y adaptando la CL para ser eficiente en tamaño, especialmente en dominios continuos. DACL aprende casos y los organiza en estructuras dinámicas de clusters. DACL es capaz de adaptarse a entornos dinámicos, donde los nuevos clusters, meta-casos o prototipos de los casos, y las estructuras asociadas de indexación (árboles discriminantes, árboles k-d, etc.) se pueden formar, actualizarse, o incluso ser eliminados. DACL ofrece una posible solución para la gestión de la gran cantidad de datos generados en un dominio continuo no supervisado (flujo de datos). Además, se propone el uso de la Multiple Case Library (MCL), que es una versión estática de una DACL, con la misma estructura pero siendo definida estáticamente para ser utilizada en dominios supervisados. El trabajo de tesis propone algunas técnicas para mejorar los procesos de indexación y de recuperación. El método de indexación más importante es el algoritmo NIAR k-d tree, que mejora el tiempo de recuperación y la competencia, comparado con una CL plana y con las técnicas basadas en el uso de estrategias de árboles k-d estándar. Partial Matching Exploration (PME) technique, la técnica propuesta, explora una base de casos jerárquica con una indexación de estructura de árbol con el objetivo de no perder los casos más similares a un caso de consulta. Esta técnica no sólo permite explorar el mejor camino coincidente, sino también varios caminos parciales alternativos coincidentes. Los resultados, a través de la experimentación realizada con bases de datos supervisadas conocidas, muestran una mejora de la competencia y del tiempo de recuperación de casos similares. Además la construcción dinámica de prototipos en DACL ha sido probada en un dominio no supervisado (dominio ambiental), donde se evalúa la contaminación del aire. La tarea central de la construcción de prototipos en DACL es la implementación de un método estocástico para el aprendizaje de nuevos casos y la gestión de prototipos. Por último, todo el sistema, integrando todos los métodos propuestos en este trabajo de investigación, se ha evaluado en dominios no supervisados simulados con varias bases de datos de una manera gradual, como se procesan los flujos de datos en la vida real. Las conclusiones, a partir de los resultados experimentales, muestran que el sistema de adaptación dinámica propuesto (DACL / MCL), junto con las estrategias de indexación y de exploración, y con las políticas de aprendizaje de casos estocásticos y de meta-casos propuestas, mejora el rendimiento de los sistemas estándar de CBR tanto en dominios supervisados (MCL) como en dominios continuos no supervisados (DACL).

show abstract

Marathon Race Planning: A Case-Based Reasoning Approach

Abstract: We describe and evaluate a novel application of case-based reasoning to help marathon runners to achieve a personal best by: (a) predicting a challenging, but realistic race-time; and (b) recommending a race-plan to achieve this time.

Cited by 14 publications

References 18 publications

How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon

How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon

Recommender Systems: A Healthy Obsession

A dynamic adaptive framework for improving case-based reasoning system performance

Contact Info

Product

Resources

About