Research On The Prediction Model Of Corn Yield Based On Data Mining

My country is a big agricultural country,and agriculture is the backbone of the national economy.As we all know,the agricultural production system is a complex system with a high degree of uncertainty.These uncertainties include many factors such as fertility grade,climate and field management,which make it difficult to predict corn yield.Corn is the main plantation in Northeast China and is very important in the whole grain production.In recent years,due to the impact of climate change,extreme weather events in Jilin Province have increased significantly,and meteorological disasters have occurred more frequently,especially natural disasters such as drought,hail,floods,etc.have increased more than usual,which has caused a certain impact on agricultural production.Studies have shown that the main meteorological factors affecting corn yield include climatic conditions such as precipitation,sunshine,and temperature.And these meteorological conditions also largely determine the yield of corn.Based on this situation,this paper uses the average temperature,precipitation,sunshine duration information of five meteorological stations in Qianguo,Siping,Changchun,Yanji,Linjiang,Jilin Province from 1988 to 2017,as well as the content of nitrogen,phosphorus and potassium in chemical fertilizers,The data of the disaster area,agricultural electricity consumption,total power of agricultural machinery,effective irrigation area,and corn production and planting area data from 1988 to 2017 were studied for the impact of meteorological conditions on corn production,and analyzed by neural network modeling Study the relationship between various influencing factors and yield,construct a corn yield forecast model,and provide a theoretical basis for corn yield forecast.The main research contents of the paper are as follows:(1)Data collection.Use the nodejs-based web crawler to obtain the corresponding corn planting area and annual corn output in the agricultural chapter of the Jilin Statistical Yearbook in the past years,and obtain the monthly data set of climate data from the China Ground International Exchange Station of China Meteorological Data Network Go to 5 weather stations(Qianguo,Siping,Changchun,Yanji,Linjiang)of Jilin Province for the monthly meteorological information over the years,and use it as a data source to construct a corn production forecast data set;(2)Data preprocessing and attribute selection.According to the experimental requirements,normalize the sample data of different dimensions;for the problem of data set attribute redundancy,this paper performs the corn yield prediction data set processing operations,including data cleaning,data completion and other preprocessing operations,and applies ReliefF The attribute selection algorithm selects attributes for the corn yield prediction data set,selects factors that have a greater impact on corn yield,and divides the data into a training set and a test set for subsequent experimental model construction.(3)Model building.Based on the BP neural network model,appropriate model parameters are selected,and the data set after attribute selection is used to predict and model corn yield.After comparison,the improved methods used in this paper are superior to the C4.5 decision tree and the BP neural network without attribute selection in terms of average absolute error,relative square root error,and model accuracy.