Construction Of The Evaluation Model Of Technical Indicators Of Badminton Doubles Competition Based On Fractal Dimension Method

In the badminton doubles competition,the prediction of the winning and losing and the technical indexes that affect the winning and losing of the competition are scientifically and quantitatively evaluated,in the context of interdisciplinary research ideas,build a widely applicable badminton technical index evaluation model,put forward the new theoretical principles of fractal theory,machine learning and other perspectives,using literature text data,mathematical statistics,model verification and introducing fractal dimension algorithm to calculate weights,replacing and improving traditional mathematical statistics methods,in this study,the data of 40 badminton men's doubles,women's doubles and mixed doubles in 2015-2019 were used as the research object,the prediction model of technical index and the evaluation model of fractal dimension weight of technical index are constructed,the results show:For the outcome prediction model based on machine learning,use principal component analysis,support vector machines and other classification training learning algorithms to train the competition data,and test and demonstrate the basic model based on the 10-fold cross-validation method,obtained a basic model for predicting the outcome of the game,draw the corresponding verification correlation curve,and finally get a suitable outcome prediction model,the prediction result of the model is:(1)Men's doubles competition: For the training set,the optimal hyperparameter ? = 0.0016,the accuracy of the prediction result reaches 93%;for the test set,the optimal hyperparameter ? = 0.0008,the accuracy of the prediction result reaches 70%.(2)Women's doubles competition: For the training set,the optimal hyperparameter ? = 0.0108,the accuracy of the prediction result reaches 96.4%;for the test set,the optimal hyperparameter ? = 0.0136,the accuracy of the prediction result reaches 83%.(3)Mixed doubles matches: For the training set and test set,when the optimal hyperparameter ? = 0.0016 is obtained,the accuracy of the prediction results reaches 87.8% and 78%,respectively.Using the principle of scale distribution of Zipf's law to characterize all technical indicators,obtain the cumulative contribution frequency and order distribution corresponding to each technical indicator,then according to the mathematical logic relationship between Zipf index and fractal dimension,the fractal dimension of each technical index is obtained.by weighting the fractal dimension of each technical index,finally,the evaluation weight and ranking of each technical index are obtained,so as to more objectively reflect the difference in the impact of each technical index on the outcome of the game,the results of the model solution show that the top three technologies that have the most positive contributions to the game are:(1)In men's doubles,they are: Push the ball in the fore court to the back court,on receiving the serve to smash shot,on the front court rush shot,midfield push stroke to the back court and in the front court to pick ball technique.(2)In women's doubles,they are: Midfield push stroke to the back court,head smash over the back court,in the frontcourt drop technique,midfield kill and catch nets technique and backhand forehand kill the ball.(3)In mixed doubles,they are: Push the flapping body-hit,the midfield receives kills blocks the net,catch nets technique,push the ball in the fore court to the back court and receive and set the net.Research conclusions: In view of the characteristics of multiple variables,high difficulty,low tracking,strong randomness and many uncontrollable factors in the evaluation model of the isolated network confrontation sports,among the outcome prediction models,principal component analysis has obvious effects and advantages in data dimensionality reduction algorithms,support vector machines have unique advantages in solving the generalization and classification problems of small sample nonlinearity and other high dimensional data,they are all powerful weapons to solve the complicated,chaos,random and other nonlinear phenomena in badminton doubles.For the evaluation of technical indicators,Zipf's law is used to characterize each technical indicator,and the Zipf and fractal dimensions of each technical indicator are calculated by the function fitting method of ordinary least squares,then weighted the fractal dimension of the technical indicators and calculated the weight of the corresponding technical indicators to construct a reasonable evaluation model;Zipf's law has a better effect on the "frequency-sequence" scale distribution of technical indicators,Fractal dimension is a powerful weapon to deal with complex nonlinear relationship;The evaluation model constructed has the characteristics of practical feasibility,universality and theoretical scientificity and forward-looking for the evaluation of doubles matches,it promotes the new function orientation and division of labor coordination of technical index evaluation,and has a good prospect of ecological research.