loss of data from ltrj infolocs when using redisltraj

liz_mo

Mon, Feb 16, 2015 8:40 AM

This solution works. However I am running it on over 30 files and in a few of
the csv files setting the NAs adds an extra row of data into the ltraj
object which creates problems when appending the infolocs back again. The
code below seems to solve the issue. Though I can't figure out why the extra
row is being added in some files....

#convert to dataframe
tr1.nas<-ld(tr1.na)#na infolocs dataframe #convert to dataframe
tr1.I<-ld(tr1.interp)#interpolated dataframe #find row added
tail(tr1.I)
tail(tr1.nas)
#remove row
tr1.nas<-tr1.nas[-132798,]
#turn back to itraj object
tr1.na<-dl(tr1.nas)
#then...re run infolocs saving and append ***

Regards,
Liz

-----Reply----------------------------

Dear Liz,

I guess the problem is both technical (the dimensions of the data frames, as
you pointed out) and biological (how does information associated to given
steps make sense after interpolation?). You're suggesting to put in NAs for
the newly interpolated points, which is probably a safe approach if the data
is considered at the location level, and not at the step level.

Anyway, one issue with 'redisltraj' is that it is written in C, which makes
editing not very easy... One way to achieve what you want, I think (not
tested, you'll have to try it by yourself), would be to:

1) Put missing locations back in the trajectory using setNA:

tr1 <- setNA(tr1, tr1[[1]][1, "date"], 60, units = "sec")

This way, you will have the correct dimensions for the main df and the
infolocs for each ltraj element, but NAs for the coordinates of the missing
locs.

2) Storing the infolocs somewhere:

info <- infolocs(tr1)

3) Removing the missing locations (!) with na.omit, and interpolate the
trajectory, as you did before:

tr1.interp <- redisltraj(na.omit(tr1), 60, type = "time"))

4) Now you should be able to transfer the infolocs:

infolocs(tr1.interp) <- info

Let me know if that works!
Mathieu.

------------------original message-----------------------------------

I am using the package AdehabitatLT to interpolate tracking data.

I am working of GPS tracks from diving seabirds, devices were set to
record locations every 60s but due to the diving behaviour of the birds I
have a lot of missing location data. In order to carry out further
analyses I need a regular trajectory so I am trying to interpolate the
tracks to every 60s. I have been using the package AdehabitatLT and have
created an ltraj object for the tracks, and managed to interpolate the
data, however when the data is interpolated I lose the data stored within
the infolocs component. I'm guessing it is something to do with the
lengths of the dataframes differing. Is there any way to transfer the
infolocs data to the interpolated values replacing any missing rows/values
with NAs?

R code:
# Creation of a "ltraj" object, with metadata ("infolocs")
tr1 <- as.ltraj(xy,date = da,id=id,infolocs = birdData[,6:15])

str(tr1)

List of 1
$ :'data.frame': 4188 obs. of  10 variables:
   ..$ x        : num [1:4188] -1.66 -1.66 -1.66 -1.65 -1.66 ...
   ..$ y        : num [1:4188] 55.6 55.6 55.6 55.6 55.6 ...
   ..$ date     : POSIXct[1:4188], format: "2014-05-06 10:17:46"
"2014-05-06 10:20:34" "2014-05-06 10:21:27" "2014-05-06 10:22:20" ...
   ..$ dx       : num [1:4188] -0.000206 -0.000587 0.003507 -0.002726
0.0001 ...
   ..$ dy       : num [1:4188] -0.000828 0.00349 -0.002739 -0.000011
-0.000042 ...
   ..$ dist     : num [1:4188] 0.000853 0.003539 0.00445 0.002726 0.000108
...
   ..$ dt       : num [1:4188] 168 53 53 57 57 53 53 54 56 53 ...
   ..$ R2n      : num [1:4188] 0.00 7.28e-07 7.72e-06 7.37e-06 7.89e-09
...
   ..$ abs.angle: num [1:4188] -1.815 1.737 -0.663 -3.138 -0.398 ...
   ..$ rel.angle: num [1:4188] NA -2.73 -2.4 -2.47 2.74 ...
   ..- attr(*, "id")= chr "1465801"
   ..- attr(*, "burst")= chr "1465801"
   ..- attr(*, "infolocs")='data.frame':          4188 obs. of  10
variables:
   .. ..$ GPS_ALT  : num [1:4188] 18.61 2.82 7.12 21.22 24.39 ...
   .. ..$ GPS_SPD  : int [1:4188] 5184 29088 48780 33732 1008 504 756 396
0 0 ...
   .. ..$ GPS_CRS  : int [1:4188] 181 88 238 130 179 109 12 137 22 164 ...
   .. ..$ LOC_TYPE : int [1:4188] 0 0 0 0 0 0 0 0 0 0 ...
   .. ..$ GPS_DIST : num [1:4188] 0 93 390 376 171 ...
   .. ..$ Essential: int [1:4188] NA NA NA NA NA NA NA NA NA NA ...
   .. ..$ time_numb: num [1:4188] 0.429 0.431 0.432 0.432 0.433 ...
   .. ..$ BIRDID   : int [1:4188] 1465801 1465801 1465801 1465801 1465801
1465801 1465801 1465801 1465801 1465801 ...
   .. ..$ SEX      : Factor w/ 1 level "M": 1 1 1 1 1 1 1 1 1 1 ...
   .. ..$ SUBCOL   : Factor w/ 1 level "IF": 1 1 1 1 1 1 1 1 1 1 ...
- attr(*, "class")= chr [1:2] "ltraj" "list"
- attr(*, "typeII")= logi TRUE
- attr(*, "regular")= logi FALSE


#interpolation of one track, at 60s.
tr1.interp<- redisltraj(na.omit(tr1), 60, type = "time")
str(tr1.interp)
List of 1
$ :'data.frame': 132864 obs. of  10 variables:
   ..$ x        : num [1:132864] -1.66 -1.66 -1.66 -1.66 -1.65 ...
   ..$ y        : num [1:132864] 55.6 55.6 55.6 55.6 55.6 ...
   ..$ date     : POSIXct[1:132864], format: "2014-05-06 10:17:46"
"2014-05-06 10:18:46" "2014-05-06 10:19:46" "2014-05-06 10:20:46" ...
   ..$ dx       : num [1:132864] -7.36e-05 -7.36e-05 -1.92e-04 8.03e-04
1.01e-03 ...
   ..$ dy       : num [1:132864] -0.000296 -0.000296 0.000554 0.001718
-0.001762 ...
   ..$ dist     : num [1:132864] 0.000305 0.000305 0.000586 0.001896
0.002029 ...
   ..$ dt       : num [1:132864] 60 60 60 60 60 60 60 60 60 60 ...
   ..$ R2n      : num [1:132864] 0.00 9.29e-08 3.71e-07 1.16e-07 3.04e-06
...
   ..$ abs.angle: num [1:132864] -1.81 -1.81 1.9 1.13 -1.05 ...
   ..$ rel.angle: num [1:132864] NA 0 -2.564 -0.771 -2.185 ...
   ..- attr(*, "id")= chr "1465801"
   ..- attr(*, "burst")= chr "1465801"
   ..- attr(*, "infolocs")='data.frame':          132864 obs. of  1
variable:
   .. ..$ pkey: Factor w/ 132864 levels "1465801.2014-05-06 10:17:46",..:
1 2 3 4 5 6 7 8 9 10 ...
- attr(*, "class")= chr [1:2] "ltraj" "list"
- attr(*, "typeII")= logi TRUE
- attr(*, "regular")= logi TRUE


Any help would be much appreciated

Thanks,
Liz

--
View this message in context: http://r-sig-ecology.471788.n2.nabble.com/loss-of-data-from-ltrj-infolocs-when-using-redisltraj-tp7579304p7579319.html
Sent from the r-sig-ecology mailing list archive at Nabble.com.