时间序列数据分析与预处理

时间序列分析中常常遇到的一个问题是如何有效地过滤噪音和约简数据。本文通过修改传统的离散的傅立叶变换来过滤噪音和进行数据的约简，并尽可能保留原始时间序列的全局变化趋势。为检验该方法的有效性，本文同时提出一个新颖的数据分类算法MCC，并用该算法对股票回报率的变化进行预测，实验结果显示，用MCC算法在预处理后的数据上进行预测，其预测的命中率达到63．68％，而在原始数据上进行预测，其预测的命中率只有48．98％。显然，通过对原始数据进行噪音过滤有效地改善了预测的精度。另外，数据的约简也提高了预测算法的效率。

Time Series Data Analysis and Pre-processing

One problem in time series analysis is how to efficiently filter noise and perform reduction on the data. In this paper, we introduce a modification of the real discrete Fourier transform and its inverse transform to filter noise and perform reduction on time series data whilst preserving its global movement. To evaluate its performance we proposed a novel classification algorithm called MCC and use it as a mining algorithm to predict the sign of stock return. The experimental results show that the average hit rate of predicting stock return on the pre-processed data set is 63.68% and just 48.98% on the original data set. This means that the prediction accuracy has been significantly improved by means of the proposed data pre-processing. Moreover, data reduction also improves the efficiency of the MCC algorithm.