Analysis of fluctuation factors of healthy exercise based on machine data mining and Internet of things

Abstract

This paper designs and implements the software part of the sports health cloud platform system based on the Internet of things cloud computing and machine learning data mining algorithm. According to the overall design scheme of cloud platform, the functions of each part of cloud platform system software are designed and implemented in a modular way, so that the cloud platform can receive the user’s movement and physical sign data, and store these data from prescription display, user history data display and operation prescription generation. Firstly, the time domain and frequency domain features are comprehensively considered. The mean value, standard deviation and correlation coefficient of any two axes of acceleration modulus vector are extracted in feature selection, and the dimension reduction is combined with principal component analysis. Finally, the nearest neighbor classification algorithm is used to identify an example. Due to the low similarity between walking and other common motion states, there is no misjudgment in general. But this paper still uses FFT transform to eliminate the possibility of misjudgment. The experimental results show that the recognition accuracy of the algorithm is high. • This paper designs and implements the software part of the sports health cloud platform system based on the Internet of things cloud computing and machine learning data mining algorithm. • The cloud platform can receive the user’s movement and physical sign data, and store these data from prescription display. • The mean value, standard deviation and correlation coefficient of any two axes of acceleration modulus vector are extracted in feature selection. • This paper still uses FFT transform to eliminate the possibility of misjudgment.