Advocating for Automation:

Adapting Current Tools in Environmental Science through R

Hannah Podzorski

rstudio::conf(2022)

July 27, 2022

Why Automate?

I advocate for automation when I want to streamline the processing of data, so my team has more time to apply their technical knowledge to the problem at hand.

https://www.reddit.com/r/ProgrammerHumor/comments/f0ag3i/automation/

In addition to talking about how I use R to automate workflows. I’m also going to talk about how I reduce the activation energy needed to get my team on board.

Reducing the Activation Energy

To help reduce activation energy when going from a current workflow to an automated work flow I (1) rephase how I pitch automation, (2) start small, and (3) get a little creative.

Let’s Start with the Pitch

When I say automation I mean streamlining a workflow by reducing the amount of unnecessary human interaction. With the ultimate goal of giving my team members more time for greater value tasks.

Differences in Workflow

Reactionary Workflow

Automated Workflow

The difference between a typically workflow, or reactionary workflow, is that all work products (A-D) will need to be updated manually, while in an automated workflow updates only need to be made once.

Pros of Automation

• Reproducibility
• It can be simple!
• Saves time
• Less human interaction means less errors

Automation not only saves you time, but it also can be simple (i.e. can be done with excel) and since less manual edits are being made there are less errors.

Where to Start?

• Start small, task can be automated in the same amount of time as the original task.
• Meet team members where they are.

Tasks I find easiest to automate are calculating summary statistics with R and using ggplot to create figures, instead of using excel.

{openxlsx}

write.csv(data, "data.csv")

{openxlsx}

write.csv(data, "data.csv")

openxlsx::write.xlsx(data, "data.xlsx")

 

Data from California’s Groundwater Ambient Monitoring and Assessment Program (GAMA). Downloaded 2022-07-11.

openxlsx allows you to read, write, and edit excel files in R. I use it to save data frames into excel rather then csv. This saves a lot of time by preventing the lose of work when a csv file is saved with unsupported edits. Also, you can use this package to create formatted excel tables that can be used in reports.

More information: https://ycphs.github.io/openxlsx/

{officer}

 

library(officer)
library(rvg)

plot <- rvg::dml(ggobj = plot)

pptx <-read_pptx() %>%
  add_slide() %>%
  ph_with(plot, ph_location(left = 1.3, top = 0.4, width = 8.75, height = 6.9))

print(pptx, "./R/Fig-Example.pptx")

officer allows you to manipulation Word and PowerPoint documents. Using officer with the rvg package allow users to edit the graphics that are exported with officer.

More informatin on officer: https://davidgohel.github.io/officer/ More information on the rvg package: https://davidgohel.github.io/rvg/

{officer}

 

library(officer)

plot <- rvg::dml(ggobj = plot)

pptx <-read_pptx() %>%
  add_slide() %>%
  ph_with(plot, ph_location(left = 1.3, top = 0.4, width = 8.75, height = 6.9))

print(pptx, "./R/Fig-Example.pptx")

officer allows you to manipulation Word and PowerPoint documents. Using officer with the rvg package allow users to edit the graphics that are exported with officer.

More informatin on officer: https://davidgohel.github.io/officer/ More information on the rvg package: https://davidgohel.github.io/rvg/

{officer}

With officer you can also edit Powerpoint templates, which can be used to add figure numbers and captions or additional text.

Delayed
Gratification

Once I’ve proven myself with these small changes, I then try adding in more tasks that require more initial investment, but will save time if used multiple times.

ProUCL

• Statistical Software for Left Censored Environmental Data
  • Calculates Upper Confidence Limits (UCLs)
• Developed by the U.S Environmental Protection Agency (EPA)

 

ProUCL is used for calculating statistics on data with and without non-detect observations,as know as left censored data. This includes simple statistics like the mean and more advance statistics like the Upper Confidence Limits. It’s used frequently in environmental risk assessments where we look at the impact chemicals in the environment can have on wildlife and humans in the area.

More information: https://www.epa.gov/land-research/proucl-software

ProUCL Automation

We wrote a function in R that subset our data in to manageable chucks and then creates an input file that is passed to Mini-Mouse Macro and excuted. Once the output file is saved ProUCl is closed.

More information on Mini-Mouse Macro: https://www.turnssoft.com/mini-mouse-macro.html

ProUCL Output

Data from California’s Groundwater Ambient Monitoring and Assessment Program (GAMA). Downloaded 2022-07-11.

We also wrote code that parsed the output file produces by ProUCL, which saved a lot of copy and pasting time.

Was automation the way to go?

Pros

• Regulators are happy
• Reduces human error
• Saves time

Cons

• Stability of ProUCL
• Requires special set up

For our purposes we think the automation was successful. It saved us time, while still allowing us to use software that is considered industry standard. It is not perfect and could be improved, but work for what we need it for.

Final Thoughts

• It’s ok to start small.
• All skill sets are welcome!
• Ultimate goal is to leave more time for greater value tasks.

Questions?

Slides and code availabe at, github.com/hannahpodzorski/advocating-for-automation

 

Contact Information:

hpodzorski@gsi-net.com

twitter - @hpodz

github - hannahpodzorski