Research On Sales Forecasting Model And System Development Of Large And Medium-sized Wheeled Tractors

Agricultural machinery sales forecast is an important means for agricultural machinery enterprises to deal with market changes and adjust business strategies.In view of the market chaos of vicious competition among production enterprises in agricultural machinery industry.This paper takes wheeled tractors as the object,takes historical data as the basis,based on the analysis of tractor market development characteristics and the correlation analysis of regional macro factors,successively determines four groups of training sample data sets and input factors of prediction model,and puts forward a method based on BP Markov chain and random forest A combined prediction model suitable for predicting the short-term sales of wheeled tractors,and a sales prediction system is developed after the model passes the self-verification test,so as to provide relevant practitioners with a means to predict the change trend of tractor sales.The results show that the prediction effect of the combined prediction model is stable,and the function test of the sales volume prediction system is normal.This study is of significance to help enterprises rationally determine production plans and reasonably adjust marketing strategies.The main research contents and conclusions of this paper are as follows:(1)Market analysis of wheeled tractors.Based on the crawler program,the statistical data of key indicators in the agricultural field and the sales data of agricultural wheeled tractors are captured.After preprocessing the original data,917822 groups of effective data are obtained.Through the change process of the development power of China’s agricultural machinery market,the role of national policies in promoting the development of agricultural mechanization is analyzed,and the sales characteristics of wheeled tractors and the main factors affecting the sales are obtained,It provides a certain basis for analyzing the influence mechanism of regional macro factors on the sales of wheeled tractors.(2)Correlation analysis of regional macro factors.The sales volume of agricultural machinery products has an inseparable relationship with the national regulation and control policies and the level of regional economic development.Through the correlation analysis between the output data of large and medium-sized tractors and the macro factor data,it is found that the total output value of agriculture,forestry,animal husbandry and fishery,the per capita disposable income of rural residents and the correlation coefficient||>0.5 between the labor force of the primary industry and the annual output of large tractors belong to a strong correlation,and the number of rural population Central financial subsidies and per capita disposable income of rural residents are closely related to medium-sized tractors,which provides a theoretical basis for determining the training samples of the prediction model..(3)Prediction model construction and optimization.Taking the sales volume of agricultural wheeled tractors in Shaanxi Province in 2020,using the sales data of national wheeled tractors and regional macro factor data from 2018 to 2020,four groups of training sample data sets and prediction model input factors are determined in turn.Based on Python programming language and standard libraries such as Sklearn and Tensorflow,BP neural network prediction model based on historical sales,BP neural network prediction model based on regional macro factors and combined prediction model based on BP Markov chain and random forest are established respectively.Taking the average absolute error,average absolute percentage error and determination coefficient as evaluation indexes,the actual effects of the three groups of prediction models are compared.It is found that the of the combined prediction model is reduced by 114.87,the is reduced to 9.47%,and the determination coefficient~!is 0.9952.It has successfully passed the model self-verification test.It is proved that the combined model can effectively overcome the shortcomings of the single prediction model and further improve the sales prediction accuracy of wheeled tractors.(4)Develop sales forecasting system.Based on the above three groups of wheeled tractor prediction models,the sales prediction system of wheeled tractor is designed and developed by using Python programming language.The overall structure design of the prediction system adopts B/S mode,the system functions are deployed on the cloud server,the wireless network realizes data interaction,and the user directly accesses the client through the web browser.The web program is developed by using Python standard library Streamlit,which includes four modules:user login,personal information management,data management and sales volume prediction.It supports users to access directly on the browser side and can respond quickly to predict and analyze product sales volume.Taking the sales volume prediction of wheeled tractors in Shaanxi Province in 2021 as an example,the integrity of each functional module of the system is tested.The results show that the overall function of the prediction system is normal,and the prediction result is stable,which can be used for the sales volume prediction of agricultural machinery enterprises.