On a compound forecasting model for dynamic data sequence;

To improve the forecasting accuracy of dynamic data sequence,we analyze the merits and demerits of both BP neural network and grey prediction method,and propose a combined model based on BP neural network and grey prediction method,which can gather advantage of the two methods and make up the disadvantage of single model as well.

Firstly, based on the temporal data sequences generated from discrete-time systems, locally-accurate approximation of the underlying system dynamics is achieved.

It is difficult to tackle the problem about testing of attributes between two data sequences for grey relational analysis because the number of data sequences must be more than 2 in order to structure a grey relational space.

Several non-dimensionalization changings of single-index data sequence and their characteristics are studied in this paper.

The purpose of filtering of a data sequence is to reduce the influence of measurement errors as possible on condition that the periodicity, ageing, and other effect quantities aroused by causal quantities should be maintained.

Grey model is an exponential model,which has high precision to observation data series fiting in with exponential rule.

The index series of inhabitant consumer price of Shandong province from Jan 1993 to Aug 2000 is studied,The Multiplication Model of dynamic series is selected,we analyze the price index series which season effect and long trend effect are done away from.