Comparative Study Of Several Methods In The Grain Production Forecast

Grain production is the important component of the national economy, ?uctuationsin grain production will lead to the ?uctuation of the entire national economy and thefinancial sector. People want to know ahead of the coming period of the changes ingrain production when they do everything what they can do possible to increase grainproduction. the propose of they do is to provide the basis for scientific decision-making,so it is very necessary to study the problem of grain production to us .This paper mainly use several statistical methods in the financial system to comparethe forecast choice, and the genetic algorithm optimization of the BP neural network wasUsed in grain production to the forecast in the first time. First we introduce several onthe impact of the impact of grain production Ring factor,then we use the impact factoras the input variables; after using the method of multiple regression analysis model ,wepredict the output by using SPSS software, second we use BP neural network thought topredict the output,the results of using BP neural network is better than the results ofusing the method of multiple regression analysis model; taking into account the BP neuralnetwork has some ?aws, we quoted a genetic algorithm to optimize the weight thresholdof BP neural network , thus avoiding the BP neural network vulnerable to local optimalsolution ?aws. The results of the analysis, genetic algorithm is used to optimize the BPneural Network forecast has better results than the result of separatly using BP neuralnetwork forecasting it.The purpose of study this article is to compare the characteristics of these typesof model and forecasts results, and explore the application of these models, accordingto di?erent requirements for people to choose a suitable model provide the basis for theforecast. we can adjustment the impacted factor according to the projected results,so wewill get higher yield in the next year's actual output.