Jump to content

User talk:Scottkosty

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
(Redirected from User talk:PGScooter)

Welcome!

Hello, Scottkosty, and welcome to Wikipedia! Thank you for your contributions. I hope you like the place and decide to stay. Here are some pages that you might find helpful:

I hope you enjoy editing here and being a Wikipedian! Please sign your messages on discussion pages using four tildes (~~~~); this will automatically insert your username and the date. If you need help, check out Wikipedia:Questions, ask me on my talk page, or ask your question on this page and then place {{helpme}} before the question. Again, welcome! Khoikhoi 06:47, 13 July 2008 (UTC)[reply]

No problem. Please let me know if you have any more questions. Khoikhoi 03:52, 14 July 2008 (UTC)[reply]

stata comment

[edit]

well, duh, I'm using stata, of course >_> just that my picture of it hasn't really improved much after the first impressions (i.e., "oh god, why is that done like that", "syntax really described anywhere?", "so, there should be a programming language somewhere; did I just miss it somehow?", "this HAS to be designed by a social scientist", etc.); I for myself am quite engineering-minded (and used to programming everything from matlab to game scripting to hard-core machine-level coding) to the extent that just seeing the dependent variable (y) being drawn on the horizontal axis and independent (x) on the vertical gives me shivers. Now that's quite naturally a very personal opinion of stata, and shouldn't be in turn be taken personally by anyone. Right, so there. --Sigmundur (talk) 20:42, 2 September 2008 (UTC)[reply]

Ah, yeah Stata does have a really unique way of doing things. I enjoy working with Matlab as well. Doing regressions manually with matrices gives you much more control and understanding with what's going on. If you are engineering-minded, I would definitely recommend SAS! It has MUCH more of a learning curve than Stata, but I think you would like it. It is much more oriented towards programmers like yourself. The only thing is, it is extremely expensive. Luckily for me, my university has licenses. In my opinion, Stata's programming logic is inefficient- it is all about know a specific command or trick, whereas I feel you should be able to do anything you need to with just combinations of a few basic commands.PGScooter (talk) 18:51, 3 September 2008 (UTC)[reply]
Hmm, I think stata is much better than SAS for the computational stuff. If you know the commands, you can do much more sophisticated statistical modeling in Stata that you would dream of in SAS. MATLAB is good for the graphs, GUIs, and it is quite versatile. I enjoy using it as well. Although with my economics background I use it much less than I use the tailored statistical packages. For stats I really dont know anything better than stata, although I have not tired eveything out there. SAS is good to prepare data, specially large sets since it does not load it to ram, and it does not use too much CPU. But if you want to do fancy stuff, I'd definitely go with Stata. Also, if you like low level language, you can actually write stuff in C, compile it, and run it in Stata at a much faster speed since it does not have to be interpreted (Stata is written in C.) It really helps if you have to loop through observations. I also program in scheme, which has very little syntax, and from the basic blocks you build more complicated stuff. I guess that is the type of language PGScooter would like, and I agree when it comes to learning the programming part. But it might be quite a nightmare to program an efficient iv regression in a very simple language... Hence, at least for statistical languages, the syntax and commands, although painful to learn, are actually a necessity. The only time when I would recommend using something more "low level" is when attempting to estimate a likelihood function that is very strange and no statistical package could conceivably be tailored enough to have something helpful. What other stuff have you tried? Brusegadi (talk) 09:25, 13 September 2008 (UTC)[reply]
Hi Brusegadi, yes, I understand what you mean. But from a programmer's standpoint, when he learns a new programming language, he usually just needs to learn a few new commands and then can use the logic he already has to do whatever he wants (although without the other commands, it would not be that efficient). I do agree that if you have something that you need to do that can be done in Stata, it is usually much faster (in terms of coding) than other packages. I still claim that Stata is not necessarily the best for "fancy stuff", although I really don't know enough about it or other packages to say it with much confidence. I see what you are saying about there being special considerations for a statistical programming language. I would really like to try R. That is what I have heard is really for serious stuff. I think that is what they use, for example, at Berkeley's statistics department. Once again, I admit my ignorance on comparing stats package. And, the more I learn Stata, obviously the more I like it. But I still would like to try R and SAS more. PGScooter (talk) 15:50, 17 September 2008 (UTC)[reply]
We use Stata and SAS where I work. I can tell you, you will not do anything fancy in SAS that you cant get done in Stata in terms of stats. I have never tried R though, so I dont know how they compare. I think that in terms of seriousness it really depends on what you are doing. Most profs I knew in the Applied Math department at Brown (who were doing computational stuff) used Matlab. Matlab, like stata, has a lot of syntax, so it can be annoying at first. After reading the R page on wiki, I think I would like working with R because it says that the semantics are derived from scheme, and I enjoyed that one :) May I ask for a hint about what you are working on? At the end of the day the language chosen comes down to the task at hand! :) Brusegadi (talk) 05:33, 18 September 2008 (UTC)[reply]
I am studying economics. I have not done anything fancy with Stata yet, such as estimating a certain likelihood function, etc., so I can't comment on anything advanced yet. My quips were mainly with the programming structure. Interesting.. I would say that most profs in Applied Math use Matlab, but of those in Applied Math that are statisticians, I would say that the majority use R. I would love to try R as well, especially since it is open source and free. In any case, I was thinking of starting a Stata group at wikipedia. I do not think the length and depth of the current Stata article reflect how many people use Stata. I consider myself an amateur as far as Stata, so I don't know how much I could contribute other than being an organizer, but I think it would be interesting to have more sections, and maybe some quick additions under the "Example Stata code" on how to do common things in Stata (such as a regression). What do you think needs attention in the Stata article? Or do you think it is fine as is? Also, if you do follow through on working with R, I would love to know what you think. PGScooter (talk) 04:21, 19 September 2008 (UTC)[reply]
I spoke to a more experienced programmer today and he did tell me that you have more control over what you are doing in R. I'll let you know if I personally have time for a try. I'll also see what I can do as sample code, perhaps something simple like a regression and some tests and and two more examples; one showing how the macro language works and another an additional stats one. I have not bee devoting much time to wiki lately, I have been very busy, but keep me posted on what you decide to do. Good luck with your PhD! Brusegadi (talk) 06:15, 19 September 2008 (UTC)[reply]
Great, I will let you know as well if I get into R. Although I am sure there are many documented experiences with R, it would be nice to see the experience from Stata user. Thanks for the nice discussion. PGScooter (talk) 19:55, 19 September 2008 (UTC)[reply]