Agfdhyk

I want to get the mean of some values over a column in the following way:
Computing the mean of those values where there is no 0 ,then when finding 0, reset and get the mean of the next values until find 0 again, and so on..
My data looks like:

      len    value

        0   0.0000

        0   0.0000

        1 307.4463

        2 308.0341

        0   0.0000

        0   0.0000

        0   0.0000

        1 305.7210

        0   0.0000

Here is a sample:

      dput(sdat)

      structure(list(len = c(0, 0, 0, 0, 1, 2, 0, 0, 0, 0, 0, 0, 0, 

      0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 2, 3, 0, 0, 0, 0, 0), value = c(0, 

      0, 0, 0, 307.446310279931, 308.034096297716, 0, 0, 0, 0, 0, 0, 

      0, 0, 0, 0, 305.720985439491, 0, 0, 0, 0, 0, 0, 308.599408144233, 

      308.511240241565, 305.843728990264, 0, 0, 0, 0, 0)), .Names = c("len", 

      "value"), row.names = 250:280, class = "data.frame")

So, I want to get another column with the means over "value". E.g. mean of (307.4463,308.0341), then 305.72..and so on..

My first try is using ave:

   sdat$mv <- ave( sdat[,"value"], (sdat[,"value"])==0, 

               FUN=function(x) mean(x,na.rm=T))

So, the output I wish would be:

          len    value  mv 

           0    0.0000  0   

           0    0.0000  0

           1  307.4463  0

           2  308.0341  307.7402   # mean of the previous 2-values 

           0    0.0000  0          

           0    0.0000  0

           0    0.0000  0 

           1  305.7210  3015.7210

            ...................

But it's not working properly..I get wrong values..and if possible I'd to get in the resulting column only one value with the mean value (in the final position before the next 0) and the rest with 0 or NA should be fine .

Any idea/suggestion?

Many thanks,

You are getting wrong results because the grouping variable doesn't have diferent values for each segment of zero-not zero-etc.

A cumsum trick will produce a good grouping variable.

f <- c(0L, cumsum(abs(diff(sdat$value != 0))))

ave(sdat$value, f, FUN = mean, na.rm = TRUE)



res <- cbind(sdat, Mean = ave(sdat$value, f, FUN = mean, na.rm = TRUE))

head(res, 10)

#    len    value     Mean

#250   0   0.0000   0.0000

#251   0   0.0000   0.0000

#252   0   0.0000   0.0000

#253   0   0.0000   0.0000

#254   1 307.4463 307.7402

#255   2 308.0341 307.7402

#256   0   0.0000   0.0000

#257   0   0.0000   0.0000

#258   0   0.0000   0.0000

#259   0   0.0000   0.0000

Final cleanup.

rm(f)