Shop OBEX P1 Docs P2 Docs Learn Events
The Most Popular Language For Machine Learning and Data Science Is … — Parallax Forums

The Most Popular Language For Machine Learning and Data Science Is …

Comments

  • Heater.Heater. Posts: 21,230
    I did'd understand this: "Python is good fit for me because I have a computer science background"
  • DavidZemonDavidZemon Posts: 2,973
    I read the whole article just so I could find the sentence quoted by Heater. I'm confused too...
  • evanhevanh Posts: 15,916
    It's probably his way of saying he knows Python is slow but it is great for sketching out some sudo-code that can even be tested. That and there's lots of experimental learning using Python in the public domain already.
  • Heater.Heater. Posts: 21,230
    That's the thing. He does not say anything of the performance, ease of use, suitability for the task, whatever of Python.

    He just says: I did comp. sci. therefor Python.

    I might infer from it that is what comp. sci. grads have used mostly in their studies now a days. Like they used to use Lisp, Sceme back in the day. Then Java. Which is kind of depressing.

    Frankly, the rest of the article is similarly wafflely and useless.

    The best part is in one of the comments (by the author himself):

    "Python is slow when you use just Python. But when you use Python for machine learning you mostly use packages written in compiled languages (C, C++, Cython, even Fortran) and you get good performance."


    So there we have it. When it comes to Machine Learning and Data Science the real stuff is written in real languages like C++ and Fortran. That is where the actual computer science is.





  • MicksterMickster Posts: 2,693
    Heater. wrote: »
    That's the thing. He does not say anything of the performance, ease of use, suitability for the task, whatever of Python.

    He just says: I did comp. sci. therefor Python.

    I might infer from it that is what comp. sci. grads have used mostly in their studies now a days. Like they used to use Lisp, Sceme back in the day. Then Java. Which is kind of depressing.

    Frankly, the rest of the article is similarly wafflely and useless.

    The best part is in one of the comments (by the author himself):

    "Python is slow when you use just Python. But when you use Python for machine learning you mostly use packages written in compiled languages (C, C++, Cython, even Fortran) and you get good performance."


    So there we have it. When it comes to Machine Learning and Data Science the real stuff is written in real languages like C++ and Fortran. That is where the actual computer science is.

    Typical sheeple type....can never explain WHY slow, bloated languages are "better".
    I bet he uses an iPhone. :D
  • Heater.Heater. Posts: 21,230
    Except this is not a typical sheeple. What with having a PhD, being an IBM Distinguished Engineer, working on Machine Learning and Optimization.

    Seems he is quite happy with C++ and is only suggesting Python for the sheeple :)
  • KeithEKeithE Posts: 957
    Heater - maybe it's along the lines of Loui's awk for AI?

    https://gist.github.com/miku/e1555144ce9b79d1321b72e1dc86b261
    Most people are surprised when I tell them what language we use in our
    undergraduate AI programming class. That's understandable. We use
    GAWK. GAWK, Gnu's version of Aho, Weinberger, and Kernighan's old
    pattern scanning language isn't even viewed as a programming language by
    most people. Like PERL and TCL, most prefer to view it as a "scripting
    language." It has no objects; it is not functional; it does no built-in
    logic programming. Their surprise turns to puzzlement when I confide
    that (a) while the students are allowed to use any language they want;
    (b) with a single exception, the best work consistently results from
    those working in GAWK. (footnote: The exception was a PASCAL
    programmer who is now an NSF graduate fellow getting a Ph.D. in
    mathematics at Harvard.) Programmers in C, C++, and LISP haven't even
    been close (we have not seen work in PROLOG or JAVA).
    ...
  • Heater.Heater. Posts: 21,230
    Interesting. I know nothing of machine learning or AI but when I think of those things pattern matching comes to mind. So perhaps the pattern matching nature of Awk is a natural fit.
  • KeithEKeithE Posts: 957
    edited 2017-01-23 00:39
    That's an old essay so at the time a lot was quick iterations, and leniency. Now I'm guessing that Loui would recommend python, javascript, or similar. The www.pyimagesearch.com guy definitely likes python and is now working on an ML book and says "Hands on implementation using the Python programming language and the Keras (which is compatible with either TensorFlow or Theano) + mxnet libraries" His kickstarter is going ok, so there must be interest.

    Edited to add link:
    https://www.kickstarter.com/projects/adrianrosebrock/deep-learning-for-computer-vision-with-python-eboo
  • JasonDorieJasonDorie Posts: 1,930
    I thought the most common computer science language in use today was profanity. It certainly gets a lot of use when I'm debugging.
  • kwinnkwinn Posts: 8,697
    JasonDorie wrote: »
    I thought the most common computer science language in use today was profanity. It certainly gets a lot of use when I'm debugging.

    ROFLOL, and it's of great use in other areas like home repairs, renovations, and general tinkering.
  • evanhevanh Posts: 15,916
    Heater. wrote: »
    So there we have it. When it comes to Machine Learning and Data Science the real stuff is written in real languages like C++ and Fortran. That is where the actual computer science is.

    Lots of smilies ... I'll make the distinction between science and engineering here. The scientific stuff is often just proof of concept when it comes to coding. One then reengineers the code to a compiled language.
  • evanhevanh Posts: 15,916
    Heater, that talk you linked a while back about coding practices in Javascript to utilise Google's runtime optimiser was a good case in point. It basically defined a compiler reengineering spec from within Javascript.
  • Heater.Heater. Posts: 21,230
    Hmm...you have a link to that Javascript thing. Sounds like I should read it again.
  • evanhevanh Posts: 15,916
    searched for "youtube google talk on optimising javascript" ... and immediately recognised the V8 name -
  • TorTor Posts: 2,010
    edited 2017-01-23 12:14
    I'm surprised by the part about gawk. I have yet to see any [g]awk code that couldn't be represented better (as in shorter an more readable and easier to modify) in Perl. I don't know any feature of awk which Perl can't do.
  • Heater.Heater. Posts: 21,230
    edited 2017-01-23 13:21
    I don't know any feature of Awk that Javascript can't do.

    Seems to me JS and it's event driven programming model would be an excellent choice if you machine learning system is spanning multiple machines and dealing with a lot of data throughput. Like this:
    https://www.burakkanber.com/blog/machine-learning-in-other-languages-introduction/

  • MikeDYurMikeDYur Posts: 2,176
    kwinn wrote: »
    JasonDorie wrote: »
    I thought the most common computer science language in use today was profanity. It certainly gets a lot of use when I'm debugging.

    ROFLOL, and it's of great use in other areas like home repairs, renovations, and general tinkering.

    Don't forget automotive repairs, my pole barn should be soundproofed.
  • TorTor Posts: 2,010
    edited 2017-01-23 16:44
    @Heater
    Yes, agreed - I concluded many years ago that if there's ever a language obsoleted then that's got to be Awk. Anything Awk can do can be done better in more modern alternatives. And if anyone thinks Perl is difficult to read, they haven't read Awk. And I don't doubt for a second that Javascript can do all of it. Python can too. And Ruby. And so on. The only old tools I still use regularly, for simple stuff, is sed and tr. Well, more tr than sed (a Perl one-liner is as easy, and more capable.)
  • Heater.Heater. Posts: 21,230
    I always thought that those old Unix utilities, sed, tr, grep, awk, etc were great for using from shell scripts. They are pretty much built into any system, no installation required, light weight and fast.

    Now a days I think, what the hell with shell scripting let's just knock this up in JS, Python, Perl or whatever takes your fancy.

    That leaves some of the simpler tools nice to have for those interactive one liners.

    There is a bunch of guys around here that want to throw away all the normal Unix user land and just boot from the Linux kernel into a JS REPL.



  • TorTor Posts: 2,010
    edited 2017-01-23 17:14
    Well, grep I use of course - so much that I didn't think of mentioning it together with tr & co. Same as I don't think of mentioning that I'm breathing all the time :)
  • KeithEKeithE Posts: 957
    "awk for ai" was written in 1997 and was written after many classes had already taken place. He simply observed that the students who used it had a better experience than those using other languages, and did state: "First it must be confessed that PERL programmers can cobble together AI
    projects well, too." The "The AWK Programming Language" and FSF awk reference card make it really easy to learn or relearn. But I doubt that many people are bothering too much with awk these days. I do find it useful for simple text processing though. Often this is throwaway scripts typed on the command line.
  • TorTor Posts: 2,010
    Yeah, I do those throwaway scripts on the command line too, but with Perl.. 
    cat whatever.txt | perl -ne 'some perl regex and test goes here'
  • FFAFFA Posts: 20
    R plus Shiny (https://www.rstudio.com/products/shiny/) plus Parallax WX ESP8266 means ML on my prop.
Sign In or Register to comment.