Research On Broiler Price Forecast Based On BP Neural Network

With the rapid development of big data technology,all walks of life began to introduce big data technology and help them to grow and develop rapidly.In recent years,research on chicken prices forecast has become the hot research topic which is widely concerned by at home and abroad scholars,to predict future chicken prices,not only has high academic value,but also has great application value for breeding enterprise strategic formulation and risk control,what else,it can provide an important reference for decision-making for manager.Throughout China's chicken breeding status,existing breeding information opaque,feed and high labor costs,the impact of the disaster disease significantly and other aspects of the factors have a great impact on the production and consumption of chicken breeding,while macroeconomic slowdown and consumer confidence in the present stage of our country chicken consumption market is sluggish.Thus,who can predict the future trend of chicken prices,who will be able to understand the opportunities of the market and bring greater profit space.This paper is divided into four stages.In the first stage,the sample is pretreated and the stock index is introduced as an independent variable;In the second stage,using BP(Back Propagation)neural network,support vector machine(SVM),K nearest neighbor(KNN)and Bayes algorithm to construct the rise and fall prediction model,and the samples were processed by correlation coefficient,principal component analysis(PCA)and removing noise month,t hrough comparison,it is found that the correct rate of BP neural network prediction is highest,followed by KNN algorithm;In the third stage,using BP neural network,support vector machines(SVM)and multiple linear regression algorithm to build price forecasting model,by comparing the experimental results,the average error of the BP neural network optimized by GA is the smallest,followed by SVM;In the fourth stage,two algorithms are selected respectively for the rise and fall forecast and the price prediction according to highest accuracy,and combine them into four groups price forecast system,through the experimental comparison,it is found that the combination model based on BP neural network has the lowest average error and the best effect.The core idea of this paper is to predict the rise and fall first and then predict the price,because of the high feature similarity of same class of data sets,dividing the sample data into two categories,on the one hand,the correlation between the data in the same sample set can be greatly enhanced,on the other hand,on the basis of knowing whether the price will rise or down in the future with the greatest extent,using the corresponding price forecasting model to predict the price,it not only reduces the discreteness of the data,but also reduces the error and makes the prediction closer to reality.