Hacker News Reader

Comment by ipnon | Hacker News Reader

Is it really this easy to become a data scientist today? Can you go from 0 to 1 with online programming tutorials like the original post and fast.ai?

View on HN

It depends on who wants to hire you. For the top jobs, this is for sure not enough.

I always tell people to please make sure they know all the contents from here https://www.youtube.com/watch?v=ZK3O402wf1c&list=PL49CF3715C... (Gilbert Strang's Linear Algebra course) before they make any claims about being a data scientist. I bet I can teach a monkey to open a CSV in pandas and call .fit() in sklearn. But do the people really understand the underlying assumptions. Most self-proclaimed data scientists don't I am sure.

I still have a hard time calling myself a data scientist. And I am three years into a relevant PhD. The more I study, the less I feel I truly know.

You still need to know SQL, and algorithms are helpful unless you expect to be spoon-fed pristine datasets that are ready for analysis.

The applicant pool seems to be full of those who took an online course and whose "personal" projects consist entirely of projects ripped from fast.ai or similar. Anymore, I get seem to get more spam on LinkedIn from people looking to be hired as a data scientist at my company than I do from recruiters looking to hire me. And looking over their resumes, I can see why they need to hustle so much. Successful candidates need to know how to do more than classify pet images from the Oxford dataset.

So to answer your question...Maybe? I mean, these candidates are certainly trying much harder than I ever needed to for a job. But presumably they eventually get hired into roles with less discerning companies.

Agree. I actually found that I set up a Postgres instance to put the data of an object and use pandas only via sqlalchemy. Most of the stuff pandas does, can be achieved a lot more efficient if done right in a real database.

I also agree about the algorithms part. For my research, I look into vehicle routing problems a lot and there is no sklearn for that or something alike. Maybe an idea for a future project.

An experienced data scientist with deep knowledge in the topics you mention easily replaces 10 of those "candidates". I feel the best way to train more serious programmers and data scientists is to teach kids to program earlier in high school. Maybe make it mandatory just as math is today. Then, a lot more students may choose CS or math as a major in college.

Well, OP says [0] they're a PhD candidate, so maybe take 'non-CS majors' quite literally, in terms of where it's hyphenated.

0 - https://news.ycombinator.com/item?id=22669091

I consider my course a thorough course. You really learn a lot of theory. This is usually what makes students at my university drop the course after two weeks of auditing. Just because the course assumes no prior CS contents (anything really, except high school math), does not mean they don't get introduced.

Oh I wasn't suggesting anything against the quality or content of your course - perhaps I misunderstood the comment I replied to, but I took it as:

> High schoolers are really just doing some web course and then getting data scientist jobs?

so my response was:

> No - or at least this isn't evidence of that - the author is a postgraduate student and the target audience appears to be other graduates/university students.

I hope there are no companies that hire a "data scientist" that only speaks HTML :)

The course aims at students of business administration. What do they usually do? Maximize some profit function. So, my background is really more traditional Operations Research.

Reply on news.ycombinator.com