摘要
本研究以中国民用汽车拥有量54年年度数据为研究材料,通过计算机模拟对比,筛选出直线方程、指数函数曲线方程、幂函数曲线方程3类6种时间序列最优动态模型。从利用其中两种模型对未来3年预测结果看,最近时间段动态模型优于总时间序列动态模型,预测时间外延年份越长误差越大而精度越低。
This report base on the Date of 54 years China Private owned Automobile quantity as sample, select the laniary function, exponential function and power function three module and six time sequence as best trends models by means of computer imitation contrast research method. The contrast result of the two modules applied in forecasting the next three years trends suggests that the more close period trends module is better than the whole 54 years historical trends modules. The report also concludes that with the longer time explanation the forecast accuracy keeps decrease.
出处
《河南科学》
2004年第6期847-849,共3页
Henan Science
关键词
拥有量
时间序列
指数函数
幂函数
预测精度
quantity
chronically
exponential function
power function
forecast accuracy
