Study On Data Mining Of The Women's Rowing Training

Rowing is an important sport of the Olympic Games. In recent years, with more and more emphasis of countries have been laid on rowing study, there has been great development. Involved in rowing study, there are fluid mechanics, materials science, exercise physiology, sports psychology, sports biomechanics, sports training etc. There are many factors that influence the rowing performance; sports training are the most important one to improve the sports ability of athletes. Among the training methods of rowing athletes, there are land training and water training, which are also divided into endurance training and speed training, upper limbs strength training and lower limbs strength training, special training and total physical coordination training. The study of these training methods and effect, on the one hand we can identify the best training method which affects training results in many of the training methods, on the other hand, we can find out the mutual influence relationship between the various training methods, and then find the best combination of these training methods in training. However, in the current study, in order to get better results, usually two training methods were quarantined for a comparative study, this research has deviated from the actual situation of the athletes training, also affected the normal arrangement of athletes training. In this study, from the actual situation of women's rowing training, we don't interrupt normal athletes training, according to the actual training data of athletes, analyzes the data from a data mining perspective, to find out the factors and the interaction law of women's rowing training. On the prediction of rowing athletes' performance, if we can predict their performance on water accurately according to different training data, when the weather is cold or not suitable for launching training under such circumstances, that accurate prediction of performance on water will be important guiding significance for the next step of the training arrangements. In addition, if we can forecast their future achievement accurately according to the training data, we can select the best athletes and eliminate the poor athletes as early as possible, which is critical to save training expense, improve training effect, and avoid delaying the prospect of athletes. In rowing training, if we can establish the automatic training plan generating system on the basis of the existing training data and time series data matching algorithm, we can effectively avoid excessive training which lead to sports injury or insufficient training of athletes that the training load is too small to reach the training requirements and improves the scientific training level as much as possible.The research of this dissertation from the following aspects:(1) Association study of the achievements and influence factors of rowing training. For many existing problems of correlation analysis to be solved, this dissertation presents a dual time sequence mining algorithm of association rules. In the first time sequence association rule mining, for dozens of different training methods, the different effects on the training performance of the relationship between them are analyzed; in the analysis of the second time sequence association, the most important training method for improving performance is fund out from numerous training methods. The results obtained are favorable for coaches to arrange the combination training between various training methods and to promote rowing athletes' coordination ability.(2) Study on prediction model of rowing athletes'performance. In order to avoid the problem of Multicollinearity that traditional forecasting methods appears. This dissertation uses the method of principal component analysis and linear regression, first, extract the principal component factor from numerous training data of rowers, and then use the linear regression method to predict the rowers'performance on water. The results show that, the method in the dissertation have very good accuracy in the prediction of 4km rowing performance on water.(3) Research on the model of the development potential of rowing athletes. At present, rower's potential predictions are mainly short-term potential assessment, and lack quantitative data. This dissertation estimates the potential of rowing athletes using the method of the curve fitting and improved AHP, the long-term potential of rowing athletes are predicted, and compared with the actual results. The improved AHP model has a good prediction results in the rowers'potential, which is important significance for early detection of elite athletes.(4) Study on time series data matching of rowing and generating automatic training plan. Through analyzing the problems faced in rowing training, this dissertation presents a method of generating automatic training plan based on time series matching algorithm. According to the characteristics of rowing, this dissertation proposes a special time series matching algorithm, SEF, which solved such time series data matching successfully by solved the common subsequence, calculated the edit distance, and combined with fuzzy matching function, and then the next training plan is generated automatically.