Hi Narragansett!

• I’m Jasmine Dumas
  • Data Scientist and useR living in Connecticut
  • Work at Simple Finance based in Portland, OR
  • Author of ttbbeer and shinyLP
• Slides and Materials available on GitHub:
  • jasdumas/talks/rhodyrstats-R-user-group

More about my background in Biomedical Engineering

• Bachelor of Science in Biomedical Engineering from the University of Hartford
  • Lots of Calculus but no Statistics courses!
• I participated in a few of research projects (one was published in the CT Medical Journal)
• I was particularly fond of computational approaches to solving engineering problems with regards to medical devices
  • Robotics
  • Aerospace
  • Computer-Aided Design
• My Engineering Path:
  • Manufacturing Engineer ➡️ R&D Intern ➡️️ R&D Engineering Technician ➡️ (…) ➡️️ Data Scientist

I learned about data science from the internet

The classic data science diagram from Drew Conway:

A updated (albeit oppinated) data science diagram from Mikhail Popov:

I'm so tired of all the hilariously wrong data science venn diagrams so I made what I think will be definitive one. pic.twitter.com/ELcPZi6atJ

— Mikhail Popov (@bearloga) March 3, 2017

“A rose by any other name would smell as sweet”

• Aside from the buzzword of “Data Science” and the sought after title “Data Scientist” people have been involved in this type of work under several different names:
  • Data Mining
  • Datalogy
  • Knowledge Discovery
  • Data Analytics
  • Business Analytics
  • Predictive Analytics
  • Machine Learning
  • Informatics
  • Applied Statistics
  • Quantitative Analysis
  • more? …

Discovering resources can be a challenge

• With many types of resources and approaches available, the decision on what to learn and who to learn it from can be overwhelming given the lack of consensus in curriculum.
  • 🎓 Graduate/Undergraduate coursework
  • 🆓 Open source coursework
  • 👾 Massively open online coursework (MOOCs)
  • 🔩 Project experience
  • 🏕 Bootcamps
  • 💼 Job experience

Dipping my toe in the water…

• Before enrolling in an expensive graduate school program, I wanted to gain some experience by combining MOOCs and project experience.
• Some considerations I had when selecting my data science curriculum:
  • low cost or free
  • reputable instructors
  • interactive platform
  • diverse programming approaches
  • applied projects for a portfolio

Asking strangers for help through cold emailing?!

• Have you ever wanted advice but didn’t know who to contact?
• Email someone in the field that you are interested in and ask about the path they took and what research they are engaged in!
• Tip on getting a response back: Make it personal!
  • I specifically emailed about advice on applying the graduate school and collaborating on a open source research project.
  

My first exposure to R programming was through shiny in 2014

• It was a mixture of learning about Bioinformatics, gene expression data, bioconductor, and designing a data product for a non-technical user with Shiny.
• Shiny is a web application framework for R
  • Some of the usual R programming headaches came up such as stringsAsFactors = FALSE
  • Gene expression data is not like the iris dataset

Sometimes unpaid volunteer work does pays off…

• Google Summer of Code (GSoC) is a global program for undergraduates and graduate students to get funding from Google to work on an open source project for the Summer!
• What it provides:
  • Opportunity to gain more programming skills
  • Interface with the open source community
  • Archival hosting of your abstract and code
  • Nobody talks about this heavily, but it does get you a lifetime referral for positions at Google!
  

Some of the results from participating in the Google Summer of Code

• A first author publication in the Bioinformatics Journal!

• A web application (currently being refactored to submit to rOpenSci)

How to make a connection to the R community

• ✅ Develop or contribute to packages (ttbbeer, shinyGEO, shinyLP)
• ✅ Answer (or ask) for help on Twitter, through the #rstats
• ✅ Complete interesting analysis projects (Web scraping and mapping breweries with import.io and R)
• ✅ Start a podcast (RTalk)
• ✅ Start a blog (jasdumas.github.io)
• ✅ Put your code on GitHub (github.com/jasdumas)
• ✅ Attend conferences (useR!)
• ✅ Volunteer to review code/papers (Mapping useRs)
• ✅ Volunteer in community initiatives (R-Ladies, Forwards, RWeekly)

Applying for jobs that utilize R

• When searching for a Data Science Role, I look for:
  • Language agnostic (R, Python or SQL focus)
  • Job description (detailed and inclusive language)
  • Opportunities for advancement (transparency and organization)
  • Professional development (provide funding for conferences or courses)
  • Diverse teams (experiences and backgrounds)
• It’s also a humbling experience to ask for help via twitter:

Does anyone want to hire me to spice up their data with science using #rstats?

I'm looking for a job! 😬

— Jasmine Dumas (@jasdumas) March 24, 2016

Every data science path is unique!

tl;dr: I love learning in a structured setting but I can learn best from doing applied data science in the workplace.

• 🎓 Began a intro grad school course in programming at DePaul University
• 🌞 Google Summer of Code 2015 Program
• ⏳Part-time Intern in a Bioinformatics Laboratory at UCONN & Data Science Group for The Hartford Insurance Company
• 💻Full-time Associate Data Scientist in a Data Science Group for The Hartford Insurance Company
• 🎓 Began intro grad school courses in Computer Science at Johns Hopkins EP and then left
• 💻Full-time Data Scientist 2 at Simple!

Starting from the ‘bottom’ is ok!

• I started with basic tutorials, books, & courses then moved on to advanced projects where I was webscraping my own data, conceiving hypotheses, and developing dashboards & reports.

Advice that I wish I heard 2.5 years ago from myself:

Ultimately, technical skills (including statistical thinking), passion, and curiosity are key attributes of a productive data scientist and essential to collaborating with others!

Alright, on to the fun stuff … R:

R packages are a great way to combine code & documentation to share with others

What even is, beer analytics?

• The use of quantitative methods to:
  • improve the production & quality of beer
  • measure customer preferences and predict seasonality
  • gather insights on industry performance

Finding beer datasets can be difficult

• Datasets: Craft Beer Sample Data, SNAP-RateBeer* - which is provided to academic researchers only
• API’s: RateBeer, Untapped, The Beer Mapping Project, BreweryDB | API’s that are available have restrictions on usage or require a user account
• Data analysis projects of beer: yhat Beer Advocate Recommender, Crowd-sourced Beer Distribution Map

Necessity breeds innovation

• I was inspired to become better at finding datasets and web scraping from the road-blocks of finding analysis-ready datasets for beer-centric projects!
• Here is a tutorial I wrote to get meta data on brewery ratings in Connecticut from Beer Advocate using Import.io

Accessing beer statistics with the ttbbeer package

• An R data package of beer statistics from U.S. Department of the Treasury, Alcohol and Tobacco Tax and Trade Bureau (TTB)
  • available on CRAN and Github

What even is, a landing page?

• Landing pages explain the web app’s usage in a clear and concise manner - Sets up users for success!
• User success can be measured if the user can:
  • Use the application as intended, contact app maintainers for assistance, see examples, or sign up for a service or account

An example of a landing home page

Make landing pages for shiny apps with shinyLP

• Adds even more Bootstrap components to make landing home pages for Shiny
  • available on CRAN and Github

The End

• Questions & Discussion!