How Do I Become an Expert in R?

Thomas Hepner

Thomas Hepner

— 3 min read
Share

June 1, 2016

1. Take the 10 Course JHU Data Science Specialization offered by Coursera:

This is an excellent place to start developing your expertise in R programming. It costs less than $500, it’s online, and is flexible with your schedule. You will cover a lot of information and many topics about the R programming language, including but not limited to:

  • Reading data, even 1GB+ data sets
  • Cleaning and transforming data
  • Utilizing prebuilt R packages
  • Developing R functions and classes
  • Retrieving data from SQL databases with the sqldf package
  • Using statistical and machine learning models for analysis with the caret package
  • Creating data visualizations with base plotting and cool packages like ggplot2

Distribution of Gas Milage

Source: http://www.statmethods.net/advgraphs/images/ggplotdensity.png

  • Building interactive web interfaces with Shiny

CDC Weekly Case Count Shiny application

Source: https://gallery.shinyapps.io/CDCPlot/

2. Seek out projects at work where you can use R programming (even if you are not in a technical role):

Excel is such a widespread (and useful!) tool in many workplaces that it can be challenging to develop expertise of other tools. In my last job as a Financial Analyst at Amazon, nearly all analysis, reporting, and forecasting on my team was done in Excel. In fact, I was the only person in my group who knew R programming at all! When I first joined, I made sure to tell my manager that I was proficient in R programming, and would be happy to use this knowledge to add value to the team. My managers made sure to put my skill to use! Each month, I built regression models in R to measure the efficiency of our marketing spending. In addition, I got to build several histograms for an ad-hoc analysis displaying the impact of launching a new subscription business on author royalties; my team found the histograms so valuable that they were included in a document that was eventually sent to our CEO, Jeff Bezos! Unfortunately, I may never know if he ever laid eyes on my glorious histograms. The takeaway: You don’t need to be an expert in R to make an impact.

3. Practice your skills in Data Science competitions:

You can compete in Data Science competitions on websites like Kaggle, CrowdAnalytix, and DrivenData. This is a great way to engage your competitive juices and practice using R for machine learning. R is a superb tool for these competitions. In fact, it’s used by the vast majority of competitors on Kaggle. R (or Python) is used in an overwhelming majority of winning entries.

Tools Used By Competitors

Source: https://www.kaggle.com/wiki/Software

Competing in a Kaggle competition was where I first learned to use the doParallel and foreach packages to speed up my model training in R; I was able to speed up the training time of my random forest algorithm from 2.5 hours to just 30 minutes. Still, even if you don’t feel like you want to compete, there are many knowledge competitions (no prizes are offered for these competitions) with incredible data sets to practice with. Check them out!

Kaggle:

  • Titanic Machine Learning from Disaster
  • Digit Recognizer
  • Facial KeyPoints Detection

DrivenData:

  • Warm Up: Predict Blood Donations
  • Pump It Up: Data Mining the Water Table

CrowdAnalytix:

  • Business Analytics for Beginners Using R – Part I
  • Business Analytics for Beginners Using R – Part II
  • Business Analytics for Beginners Using R – Part III

Read more