14 comments
I often talk with people who are trying to move into data science either from grad school or from a somewhat related career. One of my main recommendations to them is to create a data science portfolio (even if it just contains one project). I think for someone interviewing for a data science position who doesn’t have previous data science work experience, this becomes much more valuable than for those of us who do.
Excellent overview and your experience is highly valued. Could you give a rough estimate of how much time you had to invest until your portfolio started to show a nice return of investment ? (in this case I would consider time to be the major investment more than money, but please correct me if I am wrong!)
This is an interesting question, and I’ll make a note to give a thorough answer in a blog post later.
What I’d point out now is a few things. First of all, you’re right: I do consider the main investment to be time. Whereas I can now afford to view something as an investment (i.e. buying this can help me increase my revenue), when I first started I was reluctant to spend anything on improving my portfolio. Doing so felt like just compounding risk (i.e. both time and money). Now that the risk of the portfolio has been mitigated, I have a different perspective.
The reason I want to answer in a blog post is that there’s a few answers. The easiest answer is to measure calendar dates. But obviously things happened during that period that made the sales happen. And if I knew then what I knew now, it would have happened much faster. So I want my blog post to give both answers, in a thought out way, at the same time.
Ari, thank you for the prompt answer. I would very much like to read a follow up post with more details. Looking forward to it. Since I started my PhD I have spent countless hours learning R and doing things that are pretty much also a niche, so I am considering the possibility of creating a sort of blog-portfolio since I have already done most of the heavy lifting of learning and re-learning and optimizing my workflows. Thank you again for a very motivating post!
I just published my reply: https://arilamstein.com/blog/2016/10/26/long-r-portfolio-makes-money/. I hope it helps!
I’ve explicitly discussed R projects I’ve done in job interviews before and even pulled out my phone to show interviewers. Seemed to work alright for me but I can’t imagine they would have done the requisite Googling to find my medium page or anything. On that note, if you’re so inclined, check out this project I put together to scrape and track political news data in Virginia! josh-yazman.shinyapps.io/vpap-new-sentiments
thanks for sharing your experiences Ari. I am currently doing the ‘other approach’, projects ranging from ML in python to data visualization in R. Actually because I don’t have so much experience in the field yet and I am figuring out what is the most interesting to do. But I realize this won’t make me an expert in a specific niche, so that’s a disadvantage. Stuff to think about..
I’m glad that you enjoyed the post!
Is there a particular dataset or market that interests you (e.g. healthcare, politics)? That way you could still do a variety of techniques, while still maintaining a thematic coherence.
Yes, that’s also an option. I have a particular interest for healthcare, but because I haven’t found much (payed) projects in that market, I am doing work in other markets right now. Balancing between time investment and financial return..
I suspect that there is a lot of interesting work being done in healthcare analytics, and that you could indeed get it.
Here are some free resources to get you started:
Philip Morgan’s free “Positioning Crash Course”: https://philipmorganconsulting.com/positioning-crash-course/.
And his excellent podcast “The Consulting Pipeline Podcast”: http://consultingpipelinepodcast.com/
The trick is that the more specialized you become, the easier it is to get work. And the quality of the work you get will likely be more interesting.
Howdy Ari,
I’m responding to your comments on your experience hiring software engineers. I’ve been in the consulting business for 20 some years so I’ve been able to witness how many different organizations go about hiring. If you’ll let me simplify what I’ve seen to just 2 classes, sometimes its a numbers game and sometimes it is all about finding people that fit particularly well, both in terms of skills and work culture. For the later, every stone is turned. So publications are reviewed, including blogs, and publicly available code is reviewed. These days good candidates are very likely to have contributed to open source projects and/or published code on their own on things like GitHub.
My 2 cents,
Wat
I agree with your approach, Ari, & appreciate your insights. I have heard this applied to general business practice as well: find your niche/intersection/focus.
One thing that I would add: patience! I have been an R user for a few yrs, consider myself fairly advanced, & have just started my blog to continue expanding my skills & knowledge-base. Although I have financial/job-related goals for my blog, I am also creating a portfolio because I just want to get better at writing, better at R programming, and use statistical techniques that I might not get to use in my dayjob — in other words, flex my creative/methodological muscle.
If you can monetize it, great — but if you’re not doing it because you genuinely enjoy it, it’s gonna be a lot harder.