Identification of pattern mining algorithm for rugby league players positional groups separation based on movement patterns.

The application of pattern mining algorithms to extract movement patterns from sports big data can improve training specificity by facilitating a more granular evaluation of movement. Since movement patterns can only occur as consecutive, non-consecutive, or non-sequential, this study aimed to identify the best set of movement patterns for player movement profiling in professional rugby league and quantify the similarity among distinct movement patterns. Three pattern mining algorithms (l-length Closed Contiguous [LCCspm], Longest Common Subsequence [LCS] and AprioriClose) were used to extract patterns to profile elite rugby football league hookers (n = 22 players) and wingers (n = 28 players) match-games movements across 319 matches. Jaccard similarity score was used to quantify the similarity between algorithms' movement patterns and machine learning classification modelling identified the best algorithm's movement patterns to separate playing positions. LCCspm and LCS movement patterns shared a 0.19 Jaccard similarity score. AprioriClose movement patterns shared no significant Jaccard similarity with LCCspm (0.008) and LCS (0.009) patterns. The closed contiguous movement patterns profiled by LCCspm best-separated players into playing positions. Multi-layered Perceptron classification algorithm achieved the highest accuracy of 91.02% and precision, recall and F1 scores of 0.91 respectively. Therefore, we recommend the extraction of closed contiguous (consecutive) over non-consecutive and non-sequential movement patterns for separating groups of players.

Moving beyond velocity derivatives; using global positioning system data to extract sequential movement patterns at different levels of rugby league match-play.

This study aims to (a) quantify the movement patterns during rugby league match-play and (b) identify if differences exist by levels of competition within the movement patterns and units through the sequential movement pattern (SMP) algorithm. Global Positioning System data were analysed from three competition levels; four Super League regular (regular-SL), three Super League (semi-)Finals (final-SL) and four international rugby league (international) matches. The SMP framework extracted movement pattern data for each athlete within the dataset. Between competition levels, differences were analysed using linear discriminant analysis (LDA). Movement patterns were decomposed into their composite movement units; then Kruskal-Wallis rank-sum and Dunn post-hoc were used to show differences. The SMP algorithm found 121 movement patterns comprised mainly of "walk" and "jog" based movement units. The LDA had an accuracy score of 0.81, showing good separation between competition levels. Linear discriminant 1 and 2 explained 86% and 14% of the variance. The Kruskal-Wallis found differences between competition levels for 9 of 17 movement units. Differences were primarily present between regular-SL and international with other combinations showing less differences. Movement units which showed significant differences between competition levels were mainly composed of low velocities with mixed acceleration and turning angles. The SMP algorithm found 121 movement patterns across all levels of rugby league match-play, of which, 9 were found to show significant differences between competition levels. Of these nine, all showed significant differences present between international and domestic, whereas only four found differences present within the domestic levels. This study shows the SMP algorithm can be used to differentiate between levels of rugby league and that higher levels of competition may have greater velocity demands.Highlights This study shows that movement patterns and movement units can be used to investigate team sports through the application of the SMP frameworkOne hundred and twenty-one movement patterns were found to be present within rugby league match-play, with the walk- and jog-based movement units most prevalent. No movement pattern was unique to a single competition level.Further analysis revealed that the majority of movement units analysed had significant differences between international and domestic rugby league, whereas only four movement units (i.e. f,m,n,q) had significant differences within the two domestic rugby league levels.International rugby league had higher occurrences of the movement patterns consisting of higher velocity movement units (ie. T,S,y). This suggests that international rugby league players may need greater high velocity exposure in training.

Optimising classification in sport: a replication study using physical and technical-tactical performance indicators to classify competitive levels in rugby league match-play.

Determining key performance indicators and classifying players accurately between competitive levels is one of the classification challenges in sports analytics. A recent study applied Random Forest algorithm to identify important variables to classify rugby league players into academy and senior levels and achieved 82.0% and 67.5% accuracy for backs and forwards. However, the classification accuracy could be improved due to limitations in the existing method. Therefore, this study aimed to introduce and implement feature selection technique to identify key performance indicators in rugby league positional groups and assess the performances of six classification algorithms. Fifteen and fourteen of 157 performance indicators for backs and forwards were identified respectively as key performance indicators by the correlation-based feature selection method, with seven common indicators between the positional groups. Classification results show that models developed using the key performance indicators had improved performance for both positional groups than models developed using all performance indicators. 5-Nearest Neighbour produced the best classification accuracy for backs and forwards (accuracy = 85% and 77%) which is higher than the previous method's accuracies. When analysing classification questions in sport science, researchers are encouraged to evaluate multiple classification algorithms and a feature selection method should be considered for identifying key variables.

Clustering of match running and performance indicators to assess between- and within-playing position similarity in professional rugby league.

This study aimed to determine the similarity between and within positions in professional rugby league in terms of technical performance and match displacement. Here, the analyses were repeated on 3 different datasets which consisted of technical features only, displacement features only, and a combined dataset including both. Each dataset contained 7617 observations from the 2018 and 2019 Super League seasons, including 366 players from 11 teams. For each dataset, feature selection was initially used to rank features regarding their importance for predicting a player's position for each match. Subsets of 12, 11, and 27 features were retained for technical, displacement, and combined datasets for subsequent analyses. Hierarchical cluster analyses were then carried out on the positional means to find logical groupings. For the technical dataset, 3 clusters were found: (1) props, loose forwards, second-row, hooker; (2) halves; (3) wings, centres, fullback. For displacement, 4 clusters were found: (1) second-rows, halves; (2) wings, centres; (3) fullback; (4) props, loose forward, hooker. For the combined dataset, 3 clusters were found: (1) halves, fullback; (2) wings and centres; (3) props, loose forward, hooker, second-rows. These positional clusters can be used to standardise positional groups in research investigating either technical, displacement, or both constructs within rugby league.

Sequential movement pattern-mining (SMP) in field-based team-sport: A framework for quantifying spatiotemporal data and improve training specificity?

Athlete external load is typically quantified as volumes or discretised threshold values using distance, speed and time. A framework accounting for the movement sequences of athletes has previously been proposed using radio frequency data. This study developed a framework to identify sequential movement sequences using GPS-derived spatiotemporal data in team-sports and establish its stability. Thirteen rugby league players during one match were analysed to demonstrate the application of the framework. The framework (Sequential Movement Pattern-mining [SMP]) applies techniques to analyse i) geospatial data (i.e., decimal degree latitude and longitude), ii) determine players turning angles, iii) improve movement descriptor assignment, thus improving movement unit formation and iv) improve the classification and identification of players' frequent SMP. The SMP framework allows for sub-sequences of movement units to be condensed, removing repeated elements, which offers a novel technique for the quantification of similarities or dis-similarities between players and playing positions. The SMP framework provides a robust and stable method that allows, for the first time the analysis of GPS-derived data and identifies the frequent SMP of field-based team-sport athletes. The application of the SMP framework in practice could optimise the outcomes of training of field-based team-sport athletes by improving training specificity.

Development of an expected possession value model to analyse team attacking performances in rugby league.

This study aimed to evaluate team attacking performances in rugby league via expected possession value (EPV) models. Location data from 59,233 plays in 180 Super League matches across the 2019 Super League season were used. Six EPV models were generated using arbitrary zone sizes (EPV-308 and EPV-77) or aggregated according to the total zone value generated during a match (EPV-37, EPV-19, EPV-13 and EPV-9). Attacking sets were considered as Markov Chains, allowing the value of each zone visited to be estimated based on the outcome of the possession. The Kullback-Leibler Divergence was used to evaluate the reproducibility of the value generated from each zone (the reward distribution) by teams between matches. Decreasing the number of zones improved the reproducibility of reward distributions between matches but reduced the variation in zone values. After six previous matches, the subsequent match's zones had been visited on 95% or more occasions for EPV-19 (95±4%), EPV-13 (100±0%) and EPV-9 (100±0%). The KL Divergence values were infinity (EPV-308), 0.52±0.05 (EPV-77), 0.37±0.03 (EPV-37), 0.20±0.02 (EPV-19), 0.13±0.02 (EPV-13) and 0.10±0.02 (EPV-9). This study supports the use of EPV-19 and EPV-13, but not EPV-9 (too little variation in zone values), to evaluate team attacking performance in rugby league.

A league-wide investigation into variability of rugby league match running from 322 Super League games.

Purpose:This study investigated sources of variability in the overall and phase-specific running match characteristics in elite rugby league. Methods:Microtechnology data were collected from 11 Super League (SL) teams, across 322 competitive matches within the 2018 and 2019 seasons. Total distance, high-speed running (HSR) distance (>5.5 m·s-1), average speed, and average acceleration were assessed. Variability was determined using linear mixed models, with random intercepts specified for player, position, match, and club. Results:Large within-player coefficients of variation (CV) were found across whole match, ball-in-play, attack and defence for total distance (CV range = 24% to 35%) and HSR distance (37% to 96%), whereas small to moderate CVs (≤10%) were found for average speed and average acceleration. Similarly, there was higher between-player, -position, and -match variability in total distance and HSR distance when compared with average speed and average acceleration across all periods. All metrics were stable between-teams (≤5%), except HSR distance (16% to 18%). The transition period displayed the largest variability of all phases, especially for distance (up to 42%) and HSR distance (up to 165%). Conclusion:Absolute measures of displacement display large within-player and between-player, -position, and -match variability, yet average acceleration and average speed remain relatively stable across all match-periods.

Comparison of the Predictive Performance and Interpretability of Random Forest and Linear Models on Benchmark Data Sets.

The ability to interpret the predictions made by quantitative structure-activity relationships (QSARs) offers a number of advantages. While QSARs built using nonlinear modeling approaches, such as the popular Random Forest algorithm, might sometimes be more predictive than those built using linear modeling approaches, their predictions have been perceived as difficult to interpret. However, a growing number of approaches have been proposed for interpreting nonlinear QSAR models in general and Random Forest in particular. In the current work, we compare the performance of Random Forest to those of two widely used linear modeling approaches: linear Support Vector Machines (SVMs) (or Support Vector Regression (SVR)) and partial least-squares (PLS). We compare their performance in terms of their predictivity as well as the chemical interpretability of the predictions using novel scoring schemes for assessing heat map images of substructural contributions. We critically assess different approaches for interpreting Random Forest models as well as for obtaining predictions from the forest. We assess the models on a large number of widely employed public-domain benchmark data sets corresponding to regression and binary classification problems of relevance to hit identification and toxicology. We conclude that Random Forest typically yields comparable or possibly better predictive performance than the linear modeling approaches and that its predictions may also be interpreted in a chemically and biologically meaningful way. In contrast to earlier work looking at interpretation of nonlinear QSAR models, we directly compare two methodologically distinct approaches for interpreting Random Forest models. The approaches for interpreting Random Forest assessed in our article were implemented using open-source programs that we have made available to the community. These programs are the rfFC package ( https://r-forge.r-project.org/R/?group_id=1725 ) for the R statistical programming language and the Python program HeatMapWrapper [ https://doi.org/10.5281/zenodo.495163 ] for heat map generation.

Comparing the CORAL and Random Forest approaches for modelling the in vitro cytotoxicity of silica nanomaterials.

Nanotechnology is one of the most important technological developments of the 21st century. In silico methods to predict toxicity, such as quantitative structure-activity relationships (QSARs), promote the safe-by-design approach for the development of new materials, including nanomaterials. In this study, a set of cytotoxicity experimental data corresponding to 19 data points for silica nanomaterials were investigated, to compare the widely employed CORAL and Random Forest approaches in terms of their usefulness for developing so-called 'nano-QSAR' models. 'External' leave-one-out cross-validation (LOO) analysis was performed, to validate the two different approaches. An analysis of variable importance measures and signed feature contributions for both algorithms was undertaken, in order to interpret the models developed. CORAL showed a more pronounced difference between the average coefficient of determination (R²) for training and for LOO (0.83 and 0.65 for training and LOO, respectively), compared to Random Forest (0.87 and 0.78 without bootstrap sampling, 0.90 and 0.78 with bootstrap sampling), which may be due to overfitting. With regard to the physicochemical properties of the nanomaterials, the aspect ratio and zeta potential were found to be the two most important variables for Random Forest, and the average feature contributions calculated for the corresponding descriptors were consistent with the clear trends observed in the data set: less negative zeta potential values and lower aspect ratio values were associated with higher cytotoxicity. In contrast, CORAL failed to capture these trends.

Using Pareto points for model identification in predictive toxicology.

: Predictive toxicology is concerned with the development of models that are able to predict the toxicity of chemicals. A reliable prediction of toxic effects of chemicals in living systems is highly desirable in cosmetics, drug design or food protection to speed up the process of chemical compound discovery while reducing the need for lab tests. There is an extensive literature associated with the best practice of model generation and data integration but management and automated identification of relevant models from available collections of models is still an open problem. Currently, the decision on which model should be used for a new chemical compound is left to users. This paper intends to initiate the discussion on automated model identification. We present an algorithm, based on Pareto optimality, which mines model collections and identifies a model that offers a reliable prediction for a new chemical compound. The performance of this new approach is verified for two endpoints: IGC50 and LogP. The results show a great potential for automated model identification methods in predictive toxicology.