R做时间序列分析，decompose（）

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

decompose（）中type参数为additive时，季节指数是怎么计算的啊？我只知道长期趋势是移动平均得到的。

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：时间序列分析 Ecom pose 时间序列 comp

或者帮我看看后面两句话什么意思吧，就可以操作了
The function first determines the trend component using a moving average (if filter is NULL, a symmetric window with equal weights is used), and removes it from the time series. Then, the seasonal figure is computed by averaging, for each time unit, over all periods. The seasonal figure is then centered.

You have to apply centered moving average of order 12 (if monthly data) or 4 (if quarterly date) to get the trend-cycle pattern. Subtract the trend-cycle from the original data, the rest is seasoal component and irregular component. Seasonal component is the average of monthly data. For example, your data is a monthly one from 2009 to 2011, then the seasonal index for January is the average of January 2009, January 2010 and January 2011.

waterhorse 发表于 2011-12-25 14:32
You have to apply centered moving average of order 12 (if monthly data) or 4 (if quarterly date) to  ...

Thx,I have solved successfully thorough several attempt.

楼主是不是学Introductory time series with R呢？我也遇到了这个问题，按照3楼的说法好像也没解决，因为趋势项前后有缺失啊，比如月度数据，前6项与最后6项在求平均的时候缺失了。那么在求季节效应的时候用原来序列减去趋势项，缺失项怎么办？

zippo414 发表于 2011-12-27 01:11
楼主是不是学Introductory time series with R呢？我也遇到了这个问题，按照3楼的说法好像也没解决，因为趋 ...

Ignore 缺失项, unless your time series is too short.

waterhorse 发表于 2011-12-27 05:13
Ignore 缺失项, unless your time series is too short.

是呢，我按三楼大侠说的做出来了

jiao_taishan 发表于 2011-12-27 08:57
是呢，我按三楼大侠说的做出来了

楼主能把你代码给我发一下么？回复就行，或者加我QQ21543849

学习了，数据的长度对算法的结果肯定会有影响的