Application Of Functional Data Analysis Method In The Study Of Total Power Of Agricultural Machinery

China is a large agricultural country.Improving the information management level of agricultural mechanization is of great significance to improve agricultural productivity and the utilization rate of agricultural resources and realize the sustainable development of agricultural modernization.With the rapid development of the era of big data,the data types involved in agricultural mechanization information management have become relatively more complex.There are a large number of complex data with continuous representation characteristics.The traditional statistical analysis methods cannot analyze this kind of functional data.They need to use functional data analysis methods.Therefore,this paper mainly makes an empirical analysis on the total power data of agricultural machinery in China by using the functional data analysis method.Firstly,the fitting principles of several basis functions of functional data are introduced,and they are used to fit China’s total agricultural machinery power data respectively.The results show that the fitting effect of the B-sample basis function is better than that of other basis functions,and it is more applicable to data with or without periodic characteristics.Secondly,the k-means cluster analysis method and the functional clustering analysis method are discussed respectively,and the calculation method of piecewise weighted cosine similarity of function curves is proposed,and the empirical results show that the functionbased cluster analysis based on segmented distance-weighted cosine similarity coefficient proposed in this paper works best.Then the growth rate of total agricultural machinery power of 31 provinces and cities in China,i.e.,the first order derivative of the function curve classifies the curve.The results show that the development potential of China’s agricultural machinery can be divided into different categories according to different provinces,and the provinces and cities that are relatively backward in the development of agricultural machinery have large development space,and the state should increase its policy support for these provinces and cities.Finally,using multivariate statistical principal component analysis and functional principal component analysis to process the total agricultural machinery power data,the empirical analysis results of both methods show that there are two factors affecting the fluctuation of the mean value of total agricultural machinery power in China,among which the pulling effect of provinces and cities with larger total agricultural machinery power is the main factor.The functional data enables a deeper analysis of the high-dimensional data in the actual problem.And the k-means clustering was performed with the functional principal component score of the total power of agricultural machinery.The empirical study showed that the clustering results based on the functional principal component score were better than the traditional clustering results but inferior to the functional clustering results based on the segmented weighted cosine similarity coefficient.