Data Science Team Training 2021

Suggestions for Office Hours Topics

What would you or your team like to hear about from the coaches? What topics or methods can Kristin Bergman Kristin , Josh Levy, Bio- and health informatics consultant Josh , Patric Prado Patric , and Patrick Caneer Patrick illuminate during a monthly workshop? 

Comments & Events

Svetlana, Disaster Epidemiologist
How to take ETL from a solely manual process to a more efficient and automated workflow? What do we need to know and implement in terms of building a *system*?
Gina Leipertz
  • R Markdown reports - automated reporting 
  • k-anonymity in R - or other suppression methods that can be done systematically in R 
Trevor Bauer rmd +1
Patric Prado
Svetlana, Disaster Epidemiologist Svetlana do you have an example of types of ETL you'd like to automate? Are you integrating multiple datasets or just one gigantic one?

Gina Leipertz Gina So you would want HTML reports and not dashboards to display your data? And you would like this suppression to apply to the report?

Kim Gaetz, Epidemiology Supervisor at NC Division of Public Health Kim  could you provide an example of an API you'd like to access? 
Kim Gaetz, Epidemiology Supervisor at NC Division of Public Health
Patric Prado Patric there are two that I can give as examples:
One is the US Census data API and the other is the NIOCCS coding system.

Thank you!
Svetlana, Disaster Epidemiologist
Patric Prado Patric   We have multiple data feeds from multiple sources, most of them ad-hoc. Examples: 
  1. Deployment safety. Deployment information submitted by the deployed to MS Forms and manually transferred into EpiInfo by safety staff. Communications in Outlook email, manually tracked by date/time/person/subject. Future texting solution for daily safety check ins. Goal: create one seamless workflow and minimize manual processes.
  2. Toxicological outbreak. Laboratory data (possibly LIMS output in Excel), medical records (both electronic and paper, entered into Access or Excel), patient interview data (MS Forms), syndromic surveillance query results (unknown format at this time). Goal: create a seamless, automated workflow to bring these disparate sources together.
Patric Prado
Kim Gaetz, Epidemiology Supervisor at NC Division of Public Health Kim there are numerous R packages for census: https://rconsortium.github.io/censusguide/r-packages-all.html

and helpful examples:
https://cran.r-project.org/web/packages/censusapi/vignettes/getting-started.html

This seems to be the most up to date package:
https://cran.r-project.org/web/packages/tidycensus/tidycensus.pdf

Are you planning on using R? Sometimes these nifty packages help us connect, but other times we may want to query the API ourselves, like for NIOCCS: https://csams.cdc.gov/nioccs/HelpWebService.aspx

They give you what the request looks like. It usually looks like a URL. And then it spits out a JSON message which you need to parse using an R package such as jsonlite. Happy to review this if you need assistance. Who is your coach? and we can coordinate if you get stuck after you read through this material.
Patric Prado
Svetlana, Disaster Epidemiologist Svetlana I plan on throwing together a presentation on automation in two weeks time. When I do, I'll make sure to send out an invite to other coaches as well so they can invite their teams.

Thanks!
Gina Leipertz
Patric Prado Patric Hi! Yes, since we are having some issues sorting out server options to host a dashboard, we are leaning towards creating interactive RMarkdown HTML reports. We would like suppression to apply to these reports. 

Our biggest barriers for implementing RMarkdown HTML reports are getting the design to align with our organization aesthetic AND for it to be printable. 
Josh Levy, Bio- and health informatics consultant
RMDs can be knitted to either html or pdf - though to have it look like your organization you'd still need a bit of a designer's time - but you can keep reusing the design on any future reports.