Correlation analysis of performance measures for multi-label classification

In many important application domains, such as text categorization, scene classification, biomolecular analysis and medical diagnosis, examples are naturally associated with more than one class label, giving rise to multi-label classification problems. This fact has led, in recent years, to a substantial amount of research in multi-label classification. In order to evaluate and compare multi-label classifiers, researchers have adapted evaluation measures from the single-label paradigm, like Precision and Recall; and also have developed many different measures specifically for the multi-label paradigm, like Hamming Loss and Subset Accuracy. However, these evaluation measures have been used arbitrarily in multi-label classification experiments, without an objective analysis of correlation or bias. This can lead to misleading conclusions, as the experimental results may appear to favor a specific behavior depending on the subset of measures chosen. Also, as different papers in the area currently employ distinct subsets of measures, it is difficult to compare results across papers. In this work, we provide a thorough analysis of multi-label evaluation measures, and we give concrete suggestions for researchers to make an informed decision when choosing evaluation measures for multi-label classification.     

Simulated tennis matchplay in a controlled environment

The aim of this study was to develop an exercise protocol to simulate tennis matchplay on a 'category 2' surface. Match analyses were used to form the basis for the design of the protocol. The protocol involved playing against a tennis ball serving machine. Part A of the protocol comprised 92 min 46 s of simulated tennis matchplay; Part B consisted of continuous hitting to the point of 'volitional fatigue' or when the required hitting frequency for two consecutive ball feeds could no longer be maintained. Ten elite tennis players (5 males, 5 females) volunteered to participate in the study, which was performed on an indoor tennis court (Matchplay, En-Tout-Cas). Their age, body mass and estimated maximal oxygen uptake were as follows: males, 21.7 +/- 1.0 years, 73.6 +/- 2.6 kg and 58.0 +/- 1.7 ml x kg(-1) x min(-1), respectively; females, 21.9 +/- 1.3 years, 62.3 +/- 2.0 kg and 42.2 +/- 0.7 ml x kg x min(-1), respectively (mean +/- sx). Heart rate, change in body mass and time to volitional fatigue were monitored. The heart rate responses of the participants to the simulated matchplay (range: 140-157 beats x min(-1), 73-81% peak heart rate) were consistent with the results of previous studies, for 'actual' matchplay. This protocol was successful in simulating similar physiological responses in Part A to 'actual' matchplay on a 'category 2' surface, in a controlled environment; it was also a sensitive evaluation tool of skilled performance in Part B. The current protocol may be used as a baseline protocol for studying the influence of, for example, training and dietary intervention on performance.     

Physical activity and situational motivation in physical education: influence of the motivational climate and perceived ability

The influence of perceptions of the motivational climate and perceived ability on situational motivation and the physical activity behavior of 213 male and 229 female adolescent physical education students (M age = 12.56 years; SD = 0.96) was examined over a 3-day period. A significant age by gender interaction emerged, with physical activity declining from the sixth to eighth grade. The decline was more pronounced among female than male students. Perceptions of a mastery climate were strongly related to more self-determined forms of situational motivation. In contrast, perceptions of a performance climate were strongly related to less self-determined forms of situational motivation. Results of a hierarchical regression analysis revealed gender, perceived ability, and perceptions of a mastery climate to explain a significant amount of variance in physical activity. These findings suggest that promoting a mastery oriented motivational climate in physical education will foster self-determined situational motivation and physical activity.     

Motivation for physical activity in young people: entity and incremental beliefs about athletic ability

Three studies are reported of children and youth aged 11-19 years (n = 3478) examining the nature of beliefs about athletic ability. Drawing on related research in academic, moral and stereotyping domains, development of a psychometric instrument assessing athletic ability beliefs is detailed. Support was found for a multidimensional hierarchical structure that is invariant across age and gender. Confirmatory factor analysis revealed a structure comprising two higher-order factors of entity and incremental beliefs underpinned by beliefs that athletic ability is stable and a gift (entity), and is open to improvement and can be developed through learning (incremental). Incremental beliefs, indirectly through a task goal orientation, and entity beliefs directly, predicted self-reported amotivation towards physical education and sport. On the other hand, enjoyment of physical activity in youth was predicted directly by task orientation and incremental beliefs. Predictions concerning the moderating role of perceived competence were not supported. Our findings highlight the importance of ability beliefs and goals in understanding the determinants of physical activity in children and youth.