摘要
传统的模型分析方法认为经济系统信息的复杂性是由于外在的随机因素引起的 ,所以在讨论系统的确定性基础上加上随机因素 ,利用经济数学理论如随机分析理论、动态分析、统计估计、假设检验等来模拟系统的规律以达到预测的目的。而在实际对某些经济数据进行观测时发现它们的观测值带有一定的“偏倚”,传统方法中“随机干扰”期望值为零、方差不变的假设条件很少能满足 ,观测值的出现往往带有“有偏随机游动”的特点 ,如何在分析问题时 ,考虑这些“偏倚”的影响 ,这种“偏倚”有多大 ?本文通过对时间序列的分形研究 ,探讨一种新的分形分析方法 。
The traditional model method holds that the complexity of economic system information is caused by external random factors. As a result people take random factors into consideration when discussing the determination of the system, using economic mathematical theories such as random analysis, dynamic analysis, statistical estimation, hypothesis testing etc to simulate the laws of the system so as to attain the goal of prediction. However in reality, when we observe some economic data, we find that the observations more or less have some 'deviation'. Because in the traditional method, the hypothetical conditions, such as the expected value of 'random disturbance' is zero and the variance remains unchanged, can seldom be fulfilled. The observations often have the feature of 'biased random walks'. So how to take account of the influences brought by these 'deviations' in analysis, to what extent are they 'deviated'? Through a fractal study on the time series, this paper approaches a new fractal analysis method via which the information hidden in the system can be exposed completely.
