医学数据挖掘—SQL Server 2005案例分析 教学课件 作者 周怡 王世伟 主编 医学数据挖掘第4章.pptVIP

* 三、时间序列分析 Example: Number of people employed at International Systems Corporation for 20 quarters. * Quarter Five Years Ago Four Years Ago Three Years Ago Two Years Ago Last Year 1st 150 160 175 180 200 2nd 175 190 200 220 250 3rd 450 460 480 495 520 4th 140 145 160 180 190 三、时间序列分析 Example: the following data indicate the number of mergers that took place in an industry over a 19-year period. * 三、时间序列分析 Example: Number of people employed at International Systems Corporation for 20 quarters. * 三、时间序列分析 Example: the following data indicate the number of mergers that took place in an industry over a 19-year period. * 时间序列分析 * * 三、时间序列分析 Four Components of Time Series Trends 长期趋势 Seasonal variations 季节变化 Cyclical variations 周期变化 Irregular variations 不规则变化 * 三、时间序列分析 Four Components of Time Series Trends （长期趋势）：Trends in time series are the long-term movements of the series that can be characterized by steady or only slightly variable rates of change. * 三、时间序列分析 Four Components of Time Series Seasonal variations（季节变）：Seasonal variations in a time series are those variations that occur rather predictably at a particular time each year. * 三、时间序列分析 Four Components of Time Series Cyclical variations（周期变化）：Cyclical variations are movements in a time series that are recurrent but that occur in cycles of longer than a year. * * 三、时间序列分析 Four Components of Time Series Irregular variations（不规则变化）：Irregular variations constitute the class of time series movements that do not fit into the other three categories. * 三、时间序列分析 Classical Time Series Model （1）Multiplicative Time Series Model Y＝T×S×C×I （2）Additive Time Series Model Y＝T＋S＋C＋I （3）Mixed Time Series Model Y＝T×S＋C×I * 1. 趋势外推法 趋势外推法（Trend extrapolation）是根据过去和现在的发展趋势推断未来的一类方法的总称。 基本原理是回归分析，即以时间为自变量，以时间系列为因变量进行回归分析。 * 2. 移动平均法 对于平稳型时间数列可取最近n期数值的算术平均作为后期的预测值。其基本原理是：首先选择一个固定的时间周期n，将最近n期的数据加以平均作为下一期的预测值。 * 2. 移动平均法 移动平均方法具有以下的特点： 1.当时间序列的波动较大时，最近期的观察值应该赋予