Project Euler Statistics (Popularity and Effectiveness)

In this forum members can discuss topics about specific programming languages.
Post Reply
igoeja
Posts: 2
Joined: Tue Jan 11, 2011 7:01 pm

Project Euler Statistics (Popularity and Effectiveness)

Post by igoeja » Tue Jan 11, 2011 7:03 pm

Project Euler (PE) provides some statistics, but I wanted to see the effectiveness of the languages in solving problems. Although PE lists Mathematica as number 1, and on balance of popularity and effectiveness it might be, it is not the language used to solve the highest proportion of problems. For that, the language and programmers of the following are most effective.

* Frink (43%)
* PARI/GP (28%)
* Magma (21%)
* MUMPS (18%)
* Mathematica (17%)

For comparison, the results of a few other other languages

* Haskell (11%)
* Python (10%)
* Perl (10%)
* Ruby (9%)
* F# (9%)
* Scala (8%)
* C/C++ (8%)
* Java (8%)
* C# (8%)

Susanne
Posts: 32
Joined: Sun Nov 08, 2009 7:39 am

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by Susanne » Wed Jan 12, 2011 7:52 am

The statistics might not be correct. There are solvers who use several languages. After changing the language, they probably will not each time update their profile before they submit the next answer. At least I did not do so. I am sorry for this.
Image

igoeja
Posts: 2
Joined: Tue Jan 11, 2011 7:01 pm

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by igoeja » Wed Jan 12, 2011 8:29 pm

I understand. I set mine as F# because that is what I'm solving Euler problems with, but I get paid for C# and VBA. It is easy enough to imagine people working doing the reverse, setting their language as what they get paid for, but programming in another.

Conversely, the most 'successful' languages/programmers identify as using obscure languages, Frink, PARI, Magma, etc. often very math-focused. For these, it is likely indicative of the power of the languages or the people using them.

Susanne
Posts: 32
Joined: Sun Nov 08, 2009 7:39 am

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by Susanne » Thu Jan 13, 2011 9:52 am

By reading in private forums for problems I got the impression that the people who are listed in the table of Veterans or Eulerians most of the time really used the languages they set to solve the problems.
At least one user of Python complained about the slowness of Python. Solutions in Mathematica seem to be nice.
I am actually using c#, but did not set it.
Image

Aardvark_Soup
Posts: 4
Joined: Mon May 09, 2011 7:43 pm

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by Aardvark_Soup » Mon May 09, 2011 8:18 pm

Susanne wrote:The statistics might not be correct. There are solvers who use several languages. After changing the language, they probably will not each time update their profile before they submit the next answer. At least I did not do so. I am sorry for this.
I personally use Haskell for most problems but occasionally switch to D or C++. Until now, I've never considered changing the language in my profile every time I swapped. But, for the sake of statistics, I will start doing so from now.

But.. what if you're using two languages at the same time? :D This is not even a joke: I'm currently implementing the Miller-Rabin algorithm in C and planning to call it from Haskell using its Foreign Function Interface to finally solve those nasty problems involving prime tests.

Spura
Posts: 8
Joined: Mon May 16, 2011 3:49 pm

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by Spura » Tue Jun 21, 2011 8:40 am

It's just statistics and like most statistics it has to be intelligently interpreted.
Most people have less than 10 problems solved.
Most people use Java/C# etc...

So average for these languages is low. Even though Java has several user with over 300 solved, it also has about 400 users with 1 solved, and god knows how many more with less than 10 solved. Racket has better effectiveness % than Java, it has 2 users, one with 65 solved and one with 10 solved.

oliverue
Posts: 6
Joined: Wed May 25, 2011 10:16 am

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by oliverue » Fri Jun 24, 2011 9:07 am

Can someone in the know confirm that it is really true that changing the preferred language before solution submission will update statistics in favor of that language, no matter what the preferred language is later set to?

I assumed that the statistics are compiled from whatever users' current language preferences are, which would mean that all submissions of any one user go toward one language only.

If there already is per-submission accounting of language statistics, would it not be quite easy to add a language combo box for selection at submission time?

User avatar
euler
Administrator
Posts: 3068
Joined: Sun Mar 05, 2006 4:49 pm
Location: Cheshire, England
Contact:

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by euler » Sat Jun 25, 2011 7:51 am

Every hour the statistics are updated and calculations are based on the current language preference. The point is that it should reflect the language which that member feels is best suited to solving Project Euler problems in general. It is quite possible you start with one language and over time you try out a few more, but when someone gets to, say, having solved 300 problems, and has chosen language_X as his/her preferred language then they are doing that based on a lot of experience. If we allowed a per problem choice then it would not reflect necessarily the best language for that problem as much as the language he/she knew at the time they solved the problem. I appreciate that some languages are better suited to some problems than others, and some members might diligently go back and change their preferred language for problem_Y, but the idea is to capture a snapshot of the best all-round language in the view of our members.
Image
impudens simia et macrologus profundus fabulae

oliverue
Posts: 6
Joined: Wed May 25, 2011 10:16 am

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by oliverue » Thu Jul 14, 2011 8:33 pm

Thanks for the clarification, euler! Some of us would probably prefer to know how many problems where solved in what language, but your "overall best" interpretation makes sense, too, of course.

There seems to be a small bug on the Statistics page: the # of users for a language on that page doesn't always match the number of users on the language's page. The average user rating is apparently affected by this.
The 13 users counted too many for Mathematica (849 vs 836) may not change that language's user average, but RPL and COBOL are currently reported ~4% lower because of this mismatch.

arraymac
Posts: 2
Joined: Sat Sep 24, 2011 6:24 pm

Re: Project Euler Statistics (Popularity and Effectiveness)

Post by arraymac » Sat Sep 24, 2011 6:28 pm

Because I solve problems in either APL or J (K is too ragged for me.) I am glad to have the APL/J/K choice.

Post Reply