‹ Home
Exported

Data Science Team Training 2021

Jessica Arrazola, Director of Educational Strategy at CSTESvetlana, Disaster EpidemiologistJason WilkenKim Gaetz, Epidemiology Supervisor at NC Division of Public HealthBen ScottErika C BarrettCorina ChungEmma SpencerKaralee Poschman, EpidemiologistAmy LaiAriel ChristensenEdwina WilliamsJames MuncyKrystal Collier, Syndromic Surveillance Program CoordinatorDanny Kwon +95 others
Posted by Bernadette Aylward ·

Data linkage and R Studio: April office hours follow up thread

Have you used what you learned at April's office hours for your project? Do you plan to do probabilistic linkage? What about using R studio? Chime in below and connect with other learners. 

Comments & Events

Joe Tabor, Epidemiologist Joe Tabor, Epidemiologist
Interested in data cleansing (that can include probabilistic linkage) large datasets. Does probabilistic linkage use an algorithm like Soundex for both English and Spanish names?
Notified 100 people
Manuela Staneva Manuela Staneva
So far, I have taken 2 courses on familiar topics - there were OK but the level was not very high. Nothing thought provoking or engaging. I wonder if most of these courses are designed for beginners/general audiences.
Notified 100 people
Cynthia Naval Cynthia Naval
I'm able to download RStudio, but I think I'd like to run through some of the training videos at RStudio Cloud.
Notified 100 people
Britney Rust, Epidemiologist Britney Rust, Epidemiologist
Cynthia Naval Cynthia , I found a great R course on Coursera by Johns Hopkins. Here's the link: R Programming - Home | Coursera

I also found this book on R that might be helpful (It keeps things interesting with a pirate theme)  :YaRrr! The Pirate’s Guide to R (bookdown.org)

I've been slowly learning R with the help of both of those. 
Cynthia Naval Thanks, Britney!
Notified 100 people
Cynthia Naval Cynthia Naval
Thanks, Britney. All help appreciated. Finding that Python is easier, but want to learn both that and R. Especially like the idea of a pirate-themed book!
Notified 100 people
Bernadette Aylward, CSTE Bernadette Aylward
Josh Levy, Bio- and health informatics consultant Josh do you know whether Soundex can be used to link both English and Spanish names as Joe Tabor, Epidemiologist Joe  posted above? 
Notified 100 people
Kristin Bergman Kristin Bergman
Joe Tabor, Epidemiologist Joe , I know that people have used a modified version of Soundex to identify Spanish names but have not tried it myself. From what I've read, it's helpful to consider the proportion of Spanish-language names in your data set when deciding which version to use. Josh Levy, Bio- and health informatics consultant Josh or the other coaches, Patric Prado Patric and Patrick Caneer Patrick , may know more.
Notified 100 people
Josh Levy, Bio- and health informatics consultant Josh Levy, Bio- and health informatics consultant
I haven't used soundex in the context of a data linkage. I have used it for personal geneology searches on jewishgen.org. It works *ok* for eastern european names but not great...If I were you I'd just do a pubmed search on soundex with spanish names - see if there's any reviews on the subject.
Notified 100 people
Patric Prado Patric Prado
Double metaphone has some accommodations for other than english pronunciations.

Check https://cran.r-project.org/web/packages/PGRdup/PGRdup.pdf 
if you wanted to use the double metaphone function to try it out. 

I would think you can use the fuzzyjoin package or stringr to use distance metrics if you think most errors are just do to typos (I use this extensively rather than soundex/metaphone). There are differences with hyphenation within spanish names you may need to examine since most have two last names. You should probably separate names by hypen or space before examining. Here if you have any questions!
Notified 100 people