README.md

In mhpob/TelemetryR: Handle VEMCO detection data.

TelemetryR

Aggregate, separate, and disperse detections.

September 27, 2020

Since I am using data.table as my import/export/joining workhorse, dplyr and %>% was made redundant. I've removed dplyr code and pipes to reduce the amount of imported packages.

September 13, 2018

Does ACTsplit not work? Change how your input dates are written.

An update to dplyr broke the ability to select by dates. Whereas you used to be able to use a number to filter by date:

ACTsplit('C:/Users/MYPCNAME/Documents/Vemco/Vue/ReceiverLogs',
         start = 20140401, end = 20140801)

You now need to use character dates in a standard unambigious format: "yyyy-mm-dd".

ACTsplit('C:/Users/MYPCNAME/Documents/Vemco/Vue/ReceiverLogs',
         start = '2014-04-01', end = '2014-08-01')

June 12, 2018

TelemetryR Update: Now with more speed!

Hi, All- I pushed an update which allows the use of parallel computing when importing VUE csv files, a speed up when splitting detections to send out to ACT, and a more lightweight package.

Parallelization speeds things up a bit when there are a lot of detections, but probably won’t help if you’re only pulling in a file or two. It now takes almost half the time it usually does to use vemsort on our whole csv detection database (0.75 GB of data; 3.5 million detections). Note that there are now three new (optional) arguments: cl for parallel evaluation, prog_bar to see the progress of the import (with a little bit of overhead), and creation_date to download only the most-recently created files.

devtools::install_github('mhpob/TelemetryR')

Normal way, with progress bar:

system.time(
  TelemetryR::vemsort('p:/obrien/biotelemetry/detections', prog_bar = T)
)

## Reading files...
## Binding files...
## Final data manipulation...
##    user  system elapsed 
##   27.07    5.52  114.66

Parallelized, without progress bar:

cl <- parallel::makeCluster(parallel::detectCores() - 1)
system.time(
  TelemetryR::vemsort('p:/obrien/biotelemetry/detections', prog_bar = F,
                      clust = cl)
)

## Reading files...
## Binding files...
## Final data manipulation...
##    user  system elapsed 
##   11.71    3.55   67.97

parallel::stopCluster(cl)

Selecting the most-recent detections:

You can also now select files that were created after a certain date. This becomes useful if you’re only trying to analyze or send out the most-recent data–you don’t even have to touch the other stuff.

system.time(
  TelemetryR::vemsort('p:/obrien/biotelemetry/detections',
                      creation_date = '2018-01-01')
)

## Reading files...
## Binding files...
## Final data manipulation...
##    user  system elapsed 
##    2.07    2.11   16.28

Faster ACT-splitting:

ACTsplit has been redone with more-efficient loops and an increase in writing speed by using data.table::fwrite over base::write.csv. You can also pass on the parallelization/subsetting arguments to vemsort.

TelemetryR::ACTsplit('p:/obrien/biotelemetry/detections',
                     ACTtrans = 'p:/obrien/randomr/actactive.rda',
                     creation_date = '2018-01-01')

Lighter package

The dependencies on ggplot2 and pbapply have been removed, meaning that installing/updating the package should be quicker. That being said, you will have to have to install the pbapply package if you want to see a progress bar.

I hope you find this useful! As always, please feel free to contact me with any questions.

mhpob/TelemetryR documentation built on Sept. 28, 2020, 11:33 a.m.