Analysis And Forecast Of Main Influencing Factors Of Grape Production In Turpan City

Turpan grapes are more characteristic crops in Xinjiang,and Turpan is located in a low-lying basin in central Xinjiang.It is far away from the outside world.It is less affected by external disturbances and has less stable changes.The local grape industry is more stable,so Turpan grapes are selected for research Targets,analysis of impact factor factors and yield prediction analysis of the grapes in Turpan City,is conducive to promoting the healthy development of the local grape industry,is conducive to the development of sustainable ecological agriculture,and is also conducive to increasing the income of Turpan farmers and promoting the rapid economic development of Turpan And put forward strong suggestions.Because the higher the income from grape growing,the more active the farmers are in planting grapes,the greater the investment in agricultural technology and material resources in grape growing,the quality and price of grapes will increase accordingly.Therefore,the analysis of Turpan grape impact factors and yield prediction models has a positive effect on promoting the development of grape industrialization,so it is more practical to study this topic.This article is based on the relevant theoretical knowledge of R language about principal component analysis,multiple linear regression model,time series ARIMA model and MATLAB's GM(1,1)model.Based on the influencing factors of Turpan and grape production data,the data itself Data preprocessing,establishing a data evaluation system suitable for data mining,and based on this data,a principal component analysis model was established to analyze the impact factors as well as multiple linear regression models,ARIMA models and GM(1,1)models,and residual correction The GM-ARIMA model is used to predict and analyze the yield of grapes in Turpan.A comparative study of various prediction model methods is carried out to realize the effective prediction of the trend of grape yield in Turpan city.(1)Using the principal component analysis method of influencing factors to study the indicator system data,according to the analysis results,it is found that there are two main components that affect the grape production in Turpan,one is the "input component" and the other is the "natural component".The dominant variables in the first principal component are 0.89 for grape growing industry investment,0.81 human input and 0.78 fertilizer application amount.The main variables in the second principal component are average minimum temperature 0.85,temperature difference 0.79 and sunshine hours 0.74.Therefore,through analysis,we can find that the most influential factors influencing the production of grapes in Turpan are the“input components”,followed by the “natural components”.(2)Predicted by multiple linear regression models,the average absolute error(MAD)result of the prediction is 287.26,the root mean square error(RMSE)result is 1243.05,and the average absolute percentage error(MAPE)result is 1.058%.The above three error indicators are calculated The results are relatively small,indicating that the multiple linear regression model of Turpan City has higher prediction accuracy of grape production.Although the prediction accuracy of the multiple linear regression model after considering the influencing factors is high,the data for the future influencing factors are uncertain and uncontrollable.Since the influencing factors are not predicted,the future influencing factor data cannot be substituted into the regression equation to predict the future,so the future grape yield cannot be predicted.(3)The ARIMA(0,2,1)model is used to predict the yield of grapes in Turpan.The predicted value of grape yield in Turpan can reflect the trend of the true value of grape yield.The average absolute error(MAD)result of the ARIMA(0,2,1)model prediction result is 2035.59,the root mean square error(RMSE)result is 2667.77,and the average absolute percentage error(MAPE)result is 7.688%.Smaller.It showsthat the model has good prediction accuracy and fitting degree,and can be used for prediction.(4)The model was established through the GM(1,1)model modeling step and the model was tested.It was found that most of the relative errors of the grape yield prediction model in Turpan City exceeded 10%,the average relative error was 12.46%,and the average relative accuracy was 87.54 %,the model accuracy prediction level reaches level two,and the average absolute error(MAD)result from the GM(1,1)model is3333.69,the root mean square error(RMSE)result is 4098.66,and the average absolute percentage error(MAPE)The result is 12.46%,and the calculation results of the above three error indicators are large,indicating that the GM(1,1)model is not very accurate in the prediction of grape production in Turpan.Therefore,a single GM(1,1)model is used to predict the yield of grapes in Turpan.Prediction accuracy is not high.(5)The ARIMA model is used to modify the prediction residuals of the GM(1,1)model.The relative errors of the prediction results are all around 5%,and only one year exceeds 10%.The average absolute error(MAD)of the prediction results is 1741.28,the root mean square error(RMSE)result is 2275.29,and the average absolute percentage error(MAPE)result is 6.60%.The residual error modified GM-ARIMA model has higher prediction accuracy.Compared with the single GM(1,1)model,the model accuracy is higher than the single GM(1,1)model,which can be used to predict the yield of grapes in Turpan.(6)Use the residual correction GM-ARIMA model to predict the future.The real value of 2018 has been published in the Statistical Bulletin of Turpan City National Economic and Social Development,and further verified the prediction results of the residual correction GM-ARIMA model,and found that the residual correction GM-ARIMA model has high prediction accuracy and is suitable for use.According to the forecast of the yield of grapes in Turpan,it is predicted that the yield of grapes in Turpan from 2019 to2021 will be 29774.62,30621.55 and 31786.16.