Research On The Intelligent Agricultural Production System Based On Data Mining

With the rapid development of information technology,agricultural information websites provided functions such as information browsing,query and data download,but the data couldn’t be dynamic analyzed and predicted.The existed data mining tools were powerful,which were not suitable for ordinary agricultural personnel because of the formidable professional.Aiming at the problems existed in the agricultural systems,this paper constructed an intelligent agriculture production system based on data mining,which used the data mining algorithm as the research core,and realized soil fertility level analysis,yield prediction,as well as other functions.The mining processes were carried out within the system,so users could achieve the dynamic analysis of the data only by simple operations.The main research contents of this paper were as follows:(1)Based on the research of domestic and foreign present situation,this paper analyzed and summarized the problems existed in the application of data mining technology in agricultural systems,and determined the main content of the paper.The system related technologies and theories were introduced,which provided the guarantee for the realization of the system.(2)According to the characteristics of the soil grade data in the agricultural system,the pros and cons of decision tree algorithm were analyzed and the C4.5 algorithm was selected to build a classification model.Moreover,aiming at the problem of the C4.5 algorithm,an improved K-C4.5p algorithm was proposed.In view of the low efficiency of C4.5 algorithm in dealing with continuous attributes,the K-means clustering algorithm was used to separate the continuous attributes.In order to solve the problem of fitting and improve classification accuracy,the Pearson correlation coefficient was used to divide the attributes instead of the information gain rate.Experimental results showed that the K-C4.5p algorithm had a significant increase in time and accuracy.(3)In view of the characteristics of the yield data in the agricultural system,the principle of multiple regression algorithm was analyzed,and the improved residual principal component regression algorithm was proposed for the problems of multiple regression algorithm.After having extracted m principal components from the original influence factors,the residual of the first regression was put as the m+1 principal component and then the regression model was built again.The experimental results showed that the residual principal component regression model was less complex and more accurate than the traditional multiple regression model,so the improved algorithm was more suitable for yield prediction.(4)According to the design goal of the intelligent agricultural production system,the overall design and detailed design of the system were carried out,and the development of the system was completed.Through the display of the system,it was proved that the system could not only realize the basic management of agricultural production data,and be able to achieve the function of data mining by the two modules of the level of soil fertility analysis and yield prediction.Each function modules have reached the expected effect,which realized the dynamic analysis and prediction of agricultural production data.