标题：Performance analysis considering the technical-tactical variables in female judo athletes at different sport skill levels: optimization of predictors
摘要：Performance evaluation of female judokas during judo bouts is crucial for their training. The aim of this study was to 1) Compare level of achievement to relevant technical-tactical variables based on the indices of the course of fight and time spent between grip contact and technique application in senior Polish female judo athletes 2) Try to optimize sport skill level predictors using Probabilistic Neural Network (PNN). The sample of 87 Polish senior female judo athletes from all weight categories was divided into Group 1 (n=29) and Group 2 (n=58) using Polish Judo Association ranking places 1-5 and 6-23 respectively. The fights were recorded during 4 national judo tournaments. Group 1 was characterized by significantly better results expressed in indices of activity in attack,efficiency of attacks,efficiency of defense,effectiveness of attacks,and effectiveness of defense,than Group 2 (p<0.05). Moreover,Group 1 significantly predominated over Group 2 (p<0.05) in number of attacks performed in all analyzed ranges of time (1-3s,4-8s,over 8s),measured from grip contact to technique application. The optimized PNN model of fight using patterns from effectiveness of defense,efficiency of attacks,efficiency of defense,grip time to throw 4-8s showed overall accuracy prediction of 79.3% and accuracy classification for Group 1 and Group 2 of 82.3% and 77.6% respectively. The analyzed variables were able to discriminate between higher and lower performance in elite female judo athletes. Specifically,we found significant differences in analyzed variables of attack and defense actions,and time spent from grip contact to throw action according to sport skill level. These results can be useful in setting standards that should be gradually met during training for competition.
