Monkeyr

The goal of monkeyr is to package up and share the R code needed to produce flexdashboards and Rmarkdown reports from Monkeytronics monitoring data. The input data format is consistent with the Monkeytronics Sensor Node mobile app and web portal reporting functions.

Installation

Install the latest development version of monkeyr from github :

# clone the repo and `Install and Restart`
# or `devtools::load_all()`

#remotes::install_bitbucket(repo = "monkeytronics/r-analytics", subdir = "monkeyr")
devtools::install_github("monkeytronics/monkeyr", build_vignettes = TRUE)

Monkeyr is structured as pure functions. The library dependencies are listed in the DESCRIPTION file. This along with roxygen creates the NAMESPACE with all imports/exports.

Example Set Up of an Rmd File (using home report example)

We start with a Monkeytronics Sensor Node data set :

obs.csv
dev.csv
weather.csv
interv.csv

And we use the wrangling functions to create the following sanitised, wrangled data :

wrangled_obs : the joined data set
wrangled_devices : devices list minus excluded devices
devices_for_map : a list used to populate a deployment map

(If you need to copy and paste this, make sure you remove eval = FALSE!)

library(monkeyr)

wrangled_devices <- wrangle_devices(params$filtered_devices)

wrangled_weather <- wrangle_weather(params$filtered_weather)

wrangled_obs    <- wrangle_observations(
  filtered_obs   = params$filtered_obs,
  devices        = wrangled_devices,
  weather        = wrangled_weather
)

data_volume     <- get_data_volume(
  observations   = wrangled_obs,
  from_timestamp = params$fromTimeStamp,
  to_timestamp   = params$toTimeStamp
)

devices_for_map <- get_devices_for_map(
  devices        = wrangled_devices,
  data_volume    = data_volume
)

wrangled_devices<- remove_excluded_devices(
  target_data    = wrangled_devices,
  data_volume    = data_volume
)

wrangled_obs    <- remove_excluded_devices(
  target_data    = wrangled_obs,
  data_volume    = data_volume
)

Next, we can set some options for available data:

set_options(observations = wrangled_obs)

Now Run Analysis Code (Home report example)

We have everything we need to make a time series plot for humidity :

ts_chart(
  observations = wrangled_obs,
  from_timestamp = test_params()$fromTimeStamp,
  to_timestamp = test_params()$toTimeStamp,
  target_variable = "hum"
)

or temperature :

ts_chart(
  observations = wrangled_obs,
  from_timestamp = test_params()$fromTimeStamp,
  to_timestamp = test_params()$toTimeStamp,
  target_variable = "temp"
)

Test Data & Params

We want to store dummy data sets in the package to speed up testing. There are 2 sources to consider. Firstly the raw csv files which we put into dummy-data and secondly the parameters which go in dummy-params.

├── inst
│   ├── dummy-data
│   │   ├── monkey_a
│   │   |   ├── dev.csv
│   │   |   ├── interv.csv
│   │   |   ├── obs.csv
│   │   |   ├── weather.csv
│   │   ├── monkey_b
│   │   |   ├── **.csv
│   ├── dummy-params
│   │   ├── params_blank.txt
│   │   ├── params_full_map.txt
│   │   ├── params_full_2vars.txt
│   │   ├── params_full_all.txt    
│   ├── rmd
│   │   ├── home.rmd
│   │   ├── my-report.rmd

Testing your Flexdash Report in the Local Environment

During development of your flexdash report or for use with new datasets, you should put the rmd file into the inst/rmd folder. And add new data sets into inst/dummy-data/{data-set-name}/*.csv. Now you can test your code using the run_test_report function. The following parameters are needed :

run_test_report(
    report       = "home",            ## rmd file!
    dummy_data   = "monkey_a",        ## dummy data from inst/dummy-data
    dummy_params = "params_blank.txt" ## dummy params from inst/dummy-params
)

The output will be generated as a html file in html/{report}/{dummy_data}.html

Running an Analysis from inside a Docker Container

We have a function ?knit_report that takes - a list of parameters (paths to data and timestamps)

output file name (so we can keep track of requests if needed) - output directory (we need this to tell the docker container where on the host to drop the files)

We run this like so:

# the system.file calls are to access the example data installed with the package
# in production, the list of parameters will be generated by the script calling this
# report

  obs           <- "/tmp/obs.csv"
  dev           <- "/tmp/dev.csv"
  weather       <- "/tmp/weather.csv"
  interv        <- "/tmp/interv.csv"
  from          <- 1627639151
  to            <- 1630317551
  
  par_list <- monkeyr::make_params_list(
    filtered_obs           = obs,
    filtered_devices       = dev,
    filtered_weather       = weather,
    filtered_interventions = interv,
    fromTimeStamp          = from_ts,
    toTimeStamp            = to_ts
  )
  
  temporary_directory      <- Sys.getenv("TMPDIR", "/tmp")
  outfile                  <- file.path(temporary_directory, paste0("flexi", ".html"))

# knit with default file name and output dir
  monkeyr::knit_report(
    param_list = par_list,
    output_file = outfile,
    output_dir = temporary_directory,
    intermediates_dir = temporary_directory,
    envir = new.env()
  )