统计研究 ›› 2007, Vol. 24 ›› Issue (8): 17-21.

• 论文 • 上一篇    下一篇

我国统计数据质量的评估方法研究--------趋势模拟评估法及其应用

刘洪 黄燕   

  1. 中南财经政法大学信息学院
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-08-15 发布日期:2007-08-15

Research of Methods to Evaluate Quality of Statistical Data of China: Trend Simulation Evaluation Method and Its Application

LiuHong ;HuangYan   

  • Received:1900-01-01 Revised:1900-01-01 Online:2007-08-15 Published:2007-08-15

摘要: 本文采用组合模型的形式对时间序列数据的变化特点建模,在模型通过各种检验、具有良好统计预测功能的基础上,从检验异常值的角度来分析预测值与实际值之间差异的程度,找出离群数据,利用数理统计中检验实验观测数据异常值的方法,对离群数据的误差进行统计上的显著检验,从而评估统计数据的质量。文章以我国国内生产总值(GDP)为研究对象,选取我国1978-2003年间的GDP作为样本,运用趋势模拟评估法来评估我国2004年国内生产总值的准确性。对我国经济指标的时间序列数据进行了实证分析。

关键词: 统计数据质量, 趋势模拟评估法, 实证分析

Abstract: This paper constructs an assembled model to analyze time series data. Based on the model that passes all the tests and has good statistical forecasting power, it follows these steps to evaluate the quality of statistical data: firstly, from the angle of examining the outliners, find the outlying data though comparing the differences between forecasting data and actual data, then apply statistical tests to the error of outlying data using the methods of testing the observed outlying data in mathematical statistics. This paper takes the GDP of China as investigating object, selects the data from 1978 to 2003 as a sample, and uses trend simulation evaluation method to evaluate the veracity of GDP of China in 2004. It employs demonstration analysis to time series data of economic indicators of China.

 

Key words: quality of statistical data, trend simulation evaluation method, demonstration analysis