0

I started using the below link to teach myself data science, with some mathematical knowledge (including calculus, linear algebra, stats/probability) but very little programming experience:

Quora - Roman Trusov's answer on how to learn Data Science in 100 hours.

The above link has the following starting advice:

You will need an RDBMS to handle the data, so the first day would look like this:

  1. Install and configure MySQL. Import the dump into the database
  2. Read SQL basics. Spend some time doing simple exercises to get the hang of manipulating the data...

After a few days, I'm still stuck at this stage. I'd like to learn how to do data science, and was hoping that following the plan in the above link would achieve that quickly. But I'm starting to feel it is not working. Is debugging MySQL installations an important part of becoming a data scientist - so I should just get advice on that and try harder? If not, what can I do differently to achieve my goals of learning data science in similar time frame?

Community
  • 1
Henry
  • 103
  • 5
  • 5
    Anything which says Learn x in y days/hours is complete bogus. I'm 3 years into this domain, and I still think I haven't even scratched the surface, let alone learning data science. – Dawny33 Jul 02 '16 at 03:08
  • 1
    You don't know programming and the word "maths" doesn't even appear in your post so I'd say forget about becoming a data scientist in 100 hours. You have a years of work ahead. – Emre Jul 02 '16 at 03:24
  • 2
    Start with Andrew Ng's Coursera class. If that's too difficult study programming or maths as appropriate. – Emre Jul 02 '16 at 04:08
  • The question is so different to before, really you would have been better off creating a new one, rather than editing the existing one so radically. My attempt at an answer now makes absolutely no sense for the new question :-( – Neil Slater Jul 02 '16 at 16:13
  • I'm sorry :( I'll reedit again. – Henry Jul 02 '16 at 16:25
  • Thanks for the edit to be back to the original question. Marking up the edits and suggesting your own answer in the question is still maybe a bit confusing. I will make another edit, trying to keep it the same question overall as it is now, but also keeping it to Stack Exchange style as best I can. If you don't like or agree with anything I have done, you can rollback (look at the revisions) or apply your own edits to make it closer to what you intend. – Neil Slater Jul 02 '16 at 21:05
  • I hope I haven't changed your "voice" too much, and it still expresses what you wanted to ask. – Neil Slater Jul 02 '16 at 21:33

1 Answers1

9

Being good at a subject does not automatically make someone a good teacher, and it looks like Roman's answer on Quora has fallen into a trap of thinking everything he knows is simple and could be picked up quickly. Also making the potential student attempt things outside of a beginner level - effectively just by research on the web following a 2 paragraph pointer - is going to make progress slow and frustrating. Despite no doubt good intentions, the advice there is likely to give a very poor learning experience.

There are full structured tutorials available on free Massive Online Open Courses (MOOCs) available for learning data science topics. These have been put together by professionals who know how to teach, and the effort put into designing any of these courses, providing materials etc, totally dwarfs the effort that went into the advice on Quora.

Ignore the Quora answer, and sign up for one or more of these MOOCs. You can find them at Coursera, Udacity, edX and other similar places. If you want to cram a lot of study into a short time, look for "at your own pace" courses where all the materials are available to you immediately.

The almost canonical start point to test the waters would be Andrew Ng's Machine Learning course on Coursera. Machine Learning is one of the more immediately accessible and fun parts of data science, and the course goes into theory and practice with additional sections for beginners at the start covering necessary maths and programming. Following that entire course is probably around 100 hours total effort. You won't come out of it knowing data science, but you will gain useful practical skills, and get a real taster for the machine learning side of the subject.

Neil Slater
  • 28,918
  • 4
  • 80
  • 100
  • Thanks Neil. Sorry, I'm new to stackexchange and am unfamiliar with how things work. – Henry Jul 02 '16 at 16:29
  • 2
    BTW I read some of the other Quora answers, and think some of them are more useful than the 10-day lesson plan which has risen to the top there, and which IMO is flawed as I explain in my answer. The other answers are worth a look through and some of the resources linked worth bookmarking. However, following up on all of them would of course take much longer than time currently set aside. – Neil Slater Jul 02 '16 at 21:45
  • I couldn't agree more. Very nice answer ! Can't upvote twice, sorry ! – eliasah Jul 02 '16 at 21:59
  • 1
    I agree with your perspective but I think what roman tried doing was to say that do something applied because understanding in pretty much impossible unless you are very good with math but have no idea what ML is. – dksahuji Jun 07 '17 at 03:41
  • @dksahuji: Yes, that would be useful advice. But trying to follow the 10 day curriculum on Quora directly as a learning plan still looks like a daunting task for a beginner. – Neil Slater Jun 07 '17 at 07:22