Dear Helpers, I am looking for methods and tools to compare and then to concatenate two successive time series. They are both in the same frequency and they describe one phenomena. There is no time gap between them. The problem is that the method of measurements has changed between both time series and they are no statistically the same. I would like to merge them to receive one homogeneous time series. Here is an example of time series ts1 (1982-1999) and ts2 (2000-2006). Zeros are NA's.> ts1 <- structure(c(100, 48, 31, 10, 16, 25, 56, 81, 51, 45, 73, 112,103, 60, 30, 21, 10, 26, 60, 55, 40, 46, 78, 125, 55, 21, 22, 11, 13, 20, 38, 25, 17, 27, 49, 68, 40, 29, 44, 10, 21, 28, 58, 71, 50, 68, 88, 38, 71, 37, 51, 11, 30, 30, 52, 59, 40, 47, 91, 95, 62, 31, 44, 12, 27, 35, 36, 69, 32, 50, 81, 88, 72, 47, 28, 20, 9, 15, 34, 38, 10, 18, 56, 124, 79, 26, 29, 9, 16, 21, 71, 74, 52, 49, 80, 84, 52, 36, 37, 12, 21, 23, 70, 75, 69, 69, 81, 66, 53, 26, 18, 12, 12, 11, 45, 60, 40, 39, 64, 80, 72, 30, 23, 9, 18, 22, 53, 77, 45, 31, 54, 59, 26, 16, 14, 4, 4, 12, 20, 21, 7, 13, 28, 56, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 50, 36, 17, 18, 12, 28, 99, 90, 72, 69, 98, 78, 56, 31, 48, 11, 15, 31, 70, 79, 67, 80, 107, 76, 43, 15, 29, 13, 17, 30, 70, 86, 52, 35, 45, 93, 80, 49, 29, 16, 12, 23, 68, 80, 38, 52, 86, 94, 67, 30, 14, 10, 14, 21, 35, 49, 33, 20, 39, 77), .Tsp = c(1982, 1999.91666666667, 12), class = "ts")> ts2 <- structure(c(0, 0, 0, 220, 72, 45, 45, 56, 60, 80, 81, 99, 97,92, 220, 141, 85, 40, 74, 115, 88, 97, 127, 107, 90, 86, 185, 193, 100, 43, 45, 73, 74, 95, 87, 107, 77, 63, 163, 123, 106, 35, 102, 122, 103, 96, 104, 140, 135, 87, 215, 173, 56, 38, 80, 90, 113, 72, 86, 95, 88, 93, 198, 128, 40, 28, 76, 133, 80, 89, 72, 108, 102, 86, 199, 204, 84, 60, 49, 103, 121, 98, 108, 165 ), .Tsp = c(2000, 2006.91666666667, 12), class = "ts") When we plot them together the difference is visible:> ts12 <- ts(c(ts1,ts2),start=1982,frequency=12) > plot(ts12)but> plot(decompose(ts12))confirm that both describe one phenomena. I have dozens of such pairs and I would like to find a method to recalculate one ts to the other to make one homogeneous ts. Do you have any suggestions how to statistically concatenate those two ts or zoo (I tried this package) objects? Thanks in advance for the answer and the precious time you might waste. Regards, Jedrzej ---------------------------------------------------- PS. W??cz radio w Twoim komputerze! Muzyka idealna do pracy i do zabawy! - Kliknij: klik.wp.pl/?adr=http://corto.www.wp.pl/as/radiokosci.html&sid=291
It might just be a matter of scaling: ts12s <- ts(c(scale(ts1), scale(ts2)), start = 1982, frequency = 12) plot(ts12s) plot(decompose(ts12s)) 2008/3/31 J?drzej Bojanowski <jelus at wp.pl>:> Dear Helpers, > > I am looking for methods and tools to compare and then to concatenate > two successive time series. They are both in the same frequency and they > describe one phenomena. There is no time gap between them. The problem > is that the method of measurements has changed between both time series > and they are no statistically the same. I would like to merge them to > receive one homogeneous time series. > > Here is an example of time series ts1 (1982-1999) and ts2 (2000-2006). > Zeros are NA's. > > > ts1 <- structure(c(100, 48, 31, 10, 16, 25, 56, 81, 51, 45, 73, 112, > 103, 60, 30, 21, 10, 26, 60, 55, 40, 46, 78, 125, 55, 21, 22, > 11, 13, 20, 38, 25, 17, 27, 49, 68, 40, 29, 44, 10, 21, 28, 58, > 71, 50, 68, 88, 38, 71, 37, 51, 11, 30, 30, 52, 59, 40, 47, 91, > 95, 62, 31, 44, 12, 27, 35, 36, 69, 32, 50, 81, 88, 72, 47, 28, > 20, 9, 15, 34, 38, 10, 18, 56, 124, 79, 26, 29, 9, 16, 21, 71, > 74, 52, 49, 80, 84, 52, 36, 37, 12, 21, 23, 70, 75, 69, 69, 81, > 66, 53, 26, 18, 12, 12, 11, 45, 60, 40, 39, 64, 80, 72, 30, 23, > 9, 18, 22, 53, 77, 45, 31, 54, 59, 26, 16, 14, 4, 4, 12, 20, > 21, 7, 13, 28, 56, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 50, 36, > 17, 18, 12, 28, 99, 90, 72, 69, 98, 78, 56, 31, 48, 11, 15, 31, > 70, 79, 67, 80, 107, 76, 43, 15, 29, 13, 17, 30, 70, 86, 52, > 35, 45, 93, 80, 49, 29, 16, 12, 23, 68, 80, 38, 52, 86, 94, 67, > 30, 14, 10, 14, 21, 35, 49, 33, 20, 39, 77), .Tsp = c(1982, > 1999.91666666667, > 12), class = "ts") > > > ts2 <- structure(c(0, 0, 0, 220, 72, 45, 45, 56, 60, 80, 81, 99, 97, > 92, 220, 141, 85, 40, 74, 115, 88, 97, 127, 107, 90, 86, 185, > 193, 100, 43, 45, 73, 74, 95, 87, 107, 77, 63, 163, 123, 106, > 35, 102, 122, 103, 96, 104, 140, 135, 87, 215, 173, 56, 38, 80, > 90, 113, 72, 86, 95, 88, 93, 198, 128, 40, 28, 76, 133, 80, 89, > 72, 108, 102, 86, 199, 204, 84, 60, 49, 103, 121, 98, 108, 165 > ), .Tsp = c(2000, 2006.91666666667, 12), class = "ts") > > When we plot them together the difference is visible: > > > ts12 <- ts(c(ts1,ts2),start=1982,frequency=12) > > plot(ts12) > > but > > plot(decompose(ts12)) > confirm that both describe one phenomena. > > I have dozens of such pairs and I would like to find a method to > recalculate one ts to the other to make one homogeneous ts. > > Do you have any suggestions how to statistically concatenate those two > ts or zoo (I tried this package) objects? Thanks in advance for the > answer and the precious time you might waste. > > Regards, > > Jedrzej > > ---------------------------------------------------- > PS. W??cz radio w Twoim komputerze! > Muzyka idealna do pracy i do zabawy! - Kliknij: > klik.wp.pl/?adr=http://corto.www.wp.pl/as/radiokosci.html&sid=291 > > ______________________________________________ > R-help at r-project.org mailing list > stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >