[R] Help with time series

[David Winsemius]

On Sep 28, 2009, at 4:07 PM, Steve_Friedman at nps.gov wrote:

>
> Hello
>
> I'm working with a bunch of  time series data. The data are  
> downloaded from
> a server and stored as ascii files prior to reading them into R.
>
> After reading the data sets read into R with no problem and I can us  
> the ts
> function to coerce them to time series, sometimes this works and  
> sometimes
> it fails.
>
> For example.
>  P38_SubB <-
> read.table("A:\\Data\\Output\\Sparrow\\Hydro_Data\\P38_Annual.txt",
> header=TRUE,  skip=1, sep=",", stringsAsFactors=FALSE)
>             P38_SubB$GS <- rep(0.85, dim(P38_SubB)[1])  #  GS is the
> ground surface elevation, which is not included in the initial file.
>    P38_SubB$Depth <- as.numeric(P38_SubB$P38_stage) -
> as.numeric(P38_SubB$GS) # now I subtract the ground surface to  
> calculate
> water depth
> P38_SubB.ts <-  ts(data=P38_SubB, frequency = (1), start=c(1981, 1),
> end=c(2009,1)) # Here I convert to a time series
>     plot(P38_SubB.ts,ylab="Mean Annual Water Depth", xlab="Year",
> main="CSSP Subpopulation B \n  Water Depth P38")
>
> These lines work.
>
> But these do not:
>
> R3110_SubC <-
> read.table("A:\\Data\\Output\\Sparrow\\Hydro_Data\\R3110_Annual.txt",
> header=TRUE, skip=1, sep=",", stringsAsFactors=FALSE)
> R3110_SubC$GS  <- rep(5.10, dim(R3110_SubC)[1])
>   R3110_SubC$Depth <- as.numeric(R3110_SubC$R3110_stage) -
> as.numeric(R3110_SubC$GS)
> R3110_SubC.ts <- ts(data=R3110_SubC, frequency = (1), start=c(1984,  
> 1),
> end=c(2009, 1))
>
> Warning message:
> NAs introduced by coercion
>
> I am asured that the input data are the same in each input file, yet  
> some
> coerce correctly while others result in the error which prevents  
> plotting.

They probably look that same on someone's spreadsheet but could have  
some invisible character of smart-quotes or some other formatting  
anomaly. What does str show you about the types of your variables? How  
about summary or Design::describe? Do they say htat the data.frames  
have tha same structure and lack of "string anomalies".

>
> Are there any known issues with ts, what are the suggested solutions?


-- 
David Winsemius, MD
Heritage Laboratories
West Hartford, CT