Issues in Using Self-Organizing Maps in Human Movement and Sport Science

Abstract

Self-Organizing Maps (SOMs) are steadily more integrated as data-analysis tools in human movement and sport science. One of the issues limiting researchers' confidence in their applications and conclusions concerns the (arbitrary) selection of training parameters, their effect on the quality of the SOM and the sensitivity of any subsequent analyses. In this paper, we demonstrate how quality and sensitivity may be examined to increase the validity of SOM-based data-analysis. For this purpose, we use two related data sets where the research question concerns coordination variability in a volleyball spike. SOMs are an attractive tool for analysing this problem because of their ability to reduce the highdimensional time series to a two-dimensional problem while preserving the topological, non-linear relations in the original data. In a first step, we systematically search the SOM parameter space for a set of options that produces significantly lower continuity, accuracy and combined map errors and we discuss the sensitivity of SOM-based analyses of coordination variability to changes in training parameters. In a second step, we further investigate the effect of using different numbers of trials and variables on the SOM quality and sensitivity. These sensitivity analyses are able to validate the conclusions from statistical tests. Using this type of analysis can guide researchers to select SOM parameters that optimally represent their data and to examine how they affect the subsequent analyses. This may also enforce confidence in any conclusions that are drawn from studies using SOMs and enhance their integration in human movement and sport science.

Nutritional Clustering of Cookies Developed with Cocoa Shell, Soy, and Green Banana Flours Using Exploratory Methods

“…Based on these assumptions, the network chosen was a 6 × 2 hexagon, which had the DB index equal to 0.467 and the Silhouette index equal to 0.792. Also, the map chosen had quantization error (QE) equal to 0.597, topographic error (TE) equal to 0 and combined error equal to 0.840, which are measures of accuracy, continuity, and the two combined measurements, respectively (Serrien et al 2017). According to Silva et al (2019), the closer they are to zero, the better the measurement errors, although they do not have a default value.…”

Analysis of stress and academic-sports commitment through Self-organizing Artificial Neural Networks (Análisis del estrés y el compromiso académico-deportivo mediante Redes Neuronales Artificiales Auto-organizativas)

“…The process is repeated successively, until the quality indicator converges. The quality indicator used, based on the study by Kaski and Lagus (1996), includes criteria to take into account the quantification error, the level of approximation of the neuron values to the input data, the topographic error, and the possibility of viewing the data in the resulting SOM.…”

“…For the analysis of the relationship between the psychological and sociodemographic variables with the variables related to stress and anxiety, in this study a type of Artificial Intelligence (AI) called Self-organizing Artificial Neural Networks (ANN-SOM) was used, which simulate the functioning of the brain and the nervous system and are especially indicated for the treatment of large volumes of data (Serrien et al, 2017). SOM networks perform Visual Data Analysis (VDA), by means of a non-linear regression algorithm named Expectation Maximization, to graphically reflect the relationships between data without using human intervention in this process.…”

“…It was also hypothesized that the stress variables in the sports environment would be related to the same variables applied to the academic environment. Finally, we hypothesize that ANN-SOM would allow the identification of groups with psychological and sociodemographic characteristics that are very similar to each other and different from the rest of the groups, in order to study the intrinsic characteristics of each group (Serrien et al, 2017).…”

Analysis of Whole-Body Coordination Patterning in Successful and Faulty Spikes Using Self-Organising Map-Based Cluster Analysis: A Secondary Analysis

“…In a similar study investigating the proximal-to-distal coordination in young elite volleyball players using SOMs, Serrien, Goossens [ 14 ] showed that sex may be a large contributor to coordination variability, whilst maturation seemingly had no significant impact. SOMs, which are generally considered a class of artificial neural networks, are a concurrent approach being applied to investigate human movement [ 15 ]. Within the field of human movement sciences, these SOMs are adopted to explore complex movement patterns in sporting activities [ 2 , 16 ].…”