Steve Richard works as a data analyst at the Joint Office of Homeless Services (JOHS) in Multnomah County, Oregon. JOHS is an office that was created through an inter-governmental agreement between the city of Portland, Oregon and the County to allocate funds to and manage contracts for social services agencies serving the homeless population in the County.
In his role, Steve is responsible for implementing the homeless system's monitoring plan, communicating performance outcomes to various stakeholders, automating and expanding system-wide reporting and analytics capabilities, and collaborating with partners to improve data systems, all for the ultimate purpose of ending homelessness. Outside of work, Steve and his wife raise three daughters between the ages of 2 and 5. When they are not keeping him busy, he practices archery, tries to catch as much live music as possible, and occasionally sleeps.
Why did you decide to learn R?
When I started my current job, my initial charge was to write a new set of performance reports for a reporting platform we'd recently purchased. Unfortunately, the new software was plagued with technical issues. Although my office kept hearing that the platform would get off the ground soon, soon never came. To make matters worse, the older reporting platform still in use was notoriously difficult to learn and anyway, was on the road to retirement due to the aging technology it was built on.
Once this landscape was familiar to me, it became clear that I needed to look elsewhere for a reporting & analytic solution. A colleague at a previous job had introduced R to me. I was impressed by the work he'd done in it and contemplated learning it but didn't have a concrete need until I switched jobs. The reporting software woes gave me the push I needed to seek out my first R training.
How easy or difficult was learning R? What resources were most helpful?
Getting started was easier than expected. In the first month of diving into R, I attended two in-person trainings, both of which were key to familiarizing me with the RStudio interface and teaching me enough skills that I had a raft to float on. After those first two sessions, I was weeks or even months ahead of where I would've been otherwise. More importantly, after I had the foundation those trainings provided, I knew I could continue improving.
In the following months, I began using R at work. Applying what I'd learned to real-world tasks early on was a great way to build more skills and solidify what I'd already learned. Plus, I was able to start providing my stakeholders with new, useful information, which is what it's all about.
I definitely did and still do hit roadblocks, which can be frustrating. But Google is your best friend when it comes to troubleshooting R problems. Once I decide to stop being stubborn about solving a problem myself and just Google the issue I'm facing, I can generally find a solution on one of the first few sites I check. Doing this repeatedly, along with some more in-depth research, has helped me build a decent knowledge base. I'd say the important thing in the beginning is staying patient, experimenting with code, inviting failure, and using the internet to your advantage. Things eventually fall into place and when they do, it feels amazing. That moment of applying a newly discovered solution makes you feel like a total rock star for the rest of the day, which in turn makes any difficulty faced totally worthwhile.
That moment of applying a newly discovered solution makes you feel like a total rock star for the rest of the day, which in turn makes any difficulty faced totally worthwhile.
I will say that sifting through message boards was a bit challenging at first, especially before becoming familiar with different syntaxes (e.g. the base R language vs. tidyverse). Just like anything though, it gets easier with practice. Plus, everything I've written in R has contributed to a growing library of scripts (i.e. documents) that I can reference, improve upon, and poach code from. Having that library, which doesn't take all that long to build, means I don't have to remember every procedure I've ever written.
A few other key sources of info that really helped me close some gaps in my knowledge early on include the R for Data Science book and the highly responsive online community at the R for Data Science Slack channel. I've found the occasional online training very helpful as well. Beyond those sources, literally just plugging questions into a search engine has taken me far.
In what ways has learning R changed your work?
Before learning R, I was predominantly an Excel user. I'd used Tableau and PowerBI (both data visualization platforms) as well, but the foundation of much of my analytic work was Excel. Excel is great for many things and I still use it nearly every day. But like any tool, it's not great for everything. For example, much of the work I've done in Excel has required some level of manual work and/or manual documentation. This approach limited the degree of complexity I could reliably execute, increased the probability of error, and made it harder to quickly replicate analyses or reference them later.
Much of the work I've done in Excel has required some level of manual work and/or manual documentation. This approach limited the degree of complexity I could reliably execute, increased the probability of error, and made it harder to quickly replicate analyses or reference them later.
Conversely, R has enabled me to complete much more complex analyses, document procedures automatically through the process of writing them (that never gets old), replicate analyses in a fraction of the time required in Excel, access and adapt old analyses for new purposes, and increase transparency for my stakeholders. In short, R has been a total game changer. Learning it has accelerated my office's ability to learn and made me a more valuable asset in the process.
What do you think people considering learning R might not appreciate about it?
R is easier to use than it looks. Several of my colleagues sort of shudder at the sight of its command line interface. But once someone gives you the framework to put all the pieces together, it's easier than you might think to dive in.
R is also great for more than just advanced statistical or mathematical applications. I once read an article on Medium (I've since lost track of the link) in which the author wrote that R wasn't so great at data manipulation. On the contrary, my work often involves loads of data manipulation and transformation. Save for special cases, R is what I use to do that work. Roughly a year in, I've learned that the question to ask is not whether R can do something, but how.