小波分析在经济预测模型中的应用

AN APPLICATION OF WAVELET ANALYSIS ON ECONOMICAL FORECAST MODEL

经济信号也是一种时间序列 ,它和小波分析中的信号具有相同的特性 .因此 ,可将经济时间序列看成经济信号 ,应用小波进行实际经济分析和预测 .论文针对最小二乘法的不足 ,提出了多分辨回归分析处理经济数据分析的方法 .本文在建立宏观模型时 ,利用小波分析对经济数据进行预处理 ,获得能反映宏观变化趋势的低频信息 ,再用最小二乘法进行拟合和预测 ,通过对传统最小二乘法建立的模型的对比分析 ,结果表明 :本方法优于一般最小二乘法 .

Economical signals is a time-series.They have the same properties as signals in the wavelet arialysls.So,economic series can surely be treated as signals to be used in economic analysis and forecasting.Aimed to the shortcoming of Least Square Method(LSM),the paper proposed the multiresolution regression method.In order to obtain macrostructure,we used the LSM after pre-processing(Multi-scale decomposition and reconstruction) to approximate and predict.By tompared to the two methods,The result suggests that the LSM after pre-process isbetter than the LSM directed to the original data.