I work in analytic and strategy function and find it really boring: sum and divide different datasets in sql, basically sql version of basic excel. i suspect in five years so called analytic jobs would just be a basic skill required for every tech role. I have an econometrics and finance degree and have taken bunch data science course, my day job is deeply unsatisfying intellectually. I really want to prepare myself to be a data scientist or research scientist if possible. i would say i have the quantitative understanding of more advanced work but lack real project and coding experience in more quantitative tasks. i know the best way to learn is always practice and do real projects, but my everyday project and work is mostly very simple, so I most likely will just work on actual coding practice during my spare time by combing through github and mimicing what other people did and wait for the next jump. has anyone done the transition and can provide some advise? thanks a lot.
I see a lot of people wanting to do more. What they all miss is that there is nothing stopping them from just doing it as long as they have data. Your regular work won't go away overnight, but no analytics manager will ever turn down business relevant predictive information or a thorough statistical analysis that shows what BI can't. Prove your worth, and this will become your job. The team can usually hire someone else for summary numbers if your manager can show your new role adds to the bottom line. Then after a while if the old job doesn't go away to enough of your liking, leave for the job you want with the new paid experience on your resume.
Just take interviews for data scientist and see what you need to improve on to pass them. A lot of companies barely ask about your past projects anyway, it’s all about skills.
What would be the skills that come to your mind. I think a lot of questions are about stat details which can be memorized and I can understand the stat differences
Yes. There were a few things I sacrificed: pay and brand name. In exchange I got to work as a data scientist for a startup. (I’m quite young so can afford to live cheaply for a while and explore my career options.) I had a stats background and when I told people around my firm what I was interested in, someone said they knew someone looking for a junior to train up and referred me. It’s hard though, you’re essentially asking someone to take a chance on potential and that rarely rarely happens! As for questions, here are a few: - how would you model an extremely rare case? (Stratified split) - what are some ways to build a recommender system if you had no data on product purchases itself? (Similar products/customers also purchased etc) - normal stats and probability questions - difference between bagging and boosting If someone really seemed to know their stuff, I would ask them more creative questions like how would you prefer to convert categorical columns? (One hot encoding including exclusion of one column or better yet- entity embedding if we have high dimensionality)
Have you thought about getting a MS to make the switch?
I just don’t think I can learn much in school. I have the theoretical background and just need actual practice.
It wouldn't be about learning. It would be about the credential 🤷♂️