Font Size: a A A

Research On The Preparation Of The County Government Based On A Improved RBF Neural Networks

Posted on:2016-05-20Degree:MasterType:Thesis
Country:ChinaCandidate:F ZhangFull Text:PDF
GTID:2296330467495355Subject:Systems analysis and integration
Abstract/Summary:PDF Full Text Request
Today, our country is in a critical time of government institutions reforming and government functions transforming, therefore, the decision making of the institution and establishment must accommodate the framework of socialist market economy, that make the research of total amount of staffing prediction within new situation became much more meaningful. Currently, the main methods of total amount of staffing prediction are: trend analysis, regression analysis, Markov method, gray forecast method, etc. A lot of domestic experts and scholars have been study in the demain of total amount of staffing prediction, analysis the factor of political,economy, the area under jurisdiction, population, culture and morality etc. To establish the staffing model.This paper brings up an improved algorithm of RBF network algorithm, integrate the local practice, and impled this algorithm to predict the total amount of staffing of local government successfully. This paper use Shihe district, Xinyang city, Henan province as research object, and make an initial study of staffing ratify index system of local government to be the foundation of quantitative analysis and staffing total account predict. The staffing ratify index system should involved in related economic target, such as jurisdiction area, population, administrative division, GDP, industrial and agricultural output value, fiscal revenue, administrative expenses and etc. This paper use X1population size(people), X2economic level(ten thousand yuan), X3geographical area(square kilometers), X4fiscal revenue(one hundred million yuan), X5non-government organizations, X6civil servants age distribution, X7civil servants education background distribution as training sample, establish the model, try to predict the the total account of staffing in Shihe district, Xinyang city, Henan province.This paper have been innovate from these two aspect:1. Make up the blanks of total amount of staffing prediction for grass-roots local government.2. Improved the traditional RBF algorithm network model in following two aspect:(1) Optimize the evaluation of width σ. Due to the determination of function’s core width σ parameter depend on empirical study, this article introduced the GCV guidelines to optimize the width parameter σ. (2) Divided the RBF network into sub-network, and optimize computation.The outcome of the prediction method indicate that:the RBF network model based on improved RBF network algorithm is not only feasible in theory, but also provide smaller errors and better results than traditional method of total account of staffing prediction in practice.
Keywords/Search Tags:RBF, Neural network, GCV, The least square method, Total account ofstaffing, Prediction
PDF Full Text Request
Related items