ChaseDream
标题: [讨论]SAS,MATLAB,R等,用哪个好? [打印本页]
作者: yession 时间: 2007-12-5 08:25
标题: [讨论]SAS,MATLAB,R等,用哪个好?
最近头疼这件事,问了好多朋友,但是心里还是不踏实,天生对于计算机语言有种畏惧感,尽管以前C语言考试,Foxpro,Visual Basic考试我都能考很好,但是我确实比较不喜欢程序设计,讨厌电脑,不过为了做empirical不喜欢也只好学。现在想问问大家的问题是:
(1)你觉得哪个软件最好:SAS,MATLAB,R,STATA或者其他
(2)因为我只接触过eviews和R,前者好像很简单,后者感觉更难一点,但是eviews没有R强大,是不是这些程序都是越难的功能越强大?
(3)关于怎么学大家有什么建议吗?比如有什么好的电子资源能不能共享一下,或者推荐一个链接让我去下教程。
天生对于使用计算机进行数据处理没有什么天赋,总以为“勤能补拙”,可是现在如果你不掌握先进技术的话,低效率的勤依旧是不能“补拙”的
谢谢大家
作者: ProfCPA 时间: 2007-12-5 08:50
I heard that in our field, most people use SAS or Stata.
作者: yession 时间: 2007-12-5 09:38
What are you using, ProfCPA?
作者: wztt 时间: 2007-12-5 13:10
Yeah, we are more often using SAS.
I often see my senior colleages using SAS.
I am studying Matlab at the moment, SAS is not taught in my school. How about you guys?
[此贴子已经被作者于2007-12-5 13:10:21编辑过]
作者: claire123jum 时间: 2007-12-5 20:04
matlab is better.
作者: ougaihei 时间: 2007-12-6 01:23
学经济的人也会用到matlab吗?。。。。。 第一次听说。。。。
作者: ProfCPA 时间: 2007-12-6 08:57
To yession: I am learning Stata. Just find time is too limited...
For empirical research, one of Stata or SAS should be enough.. Although it is not bad to know both.
作者: claire123jum 时间: 2007-12-6 12:42
不写中文说不清楚,写个全的.
说的不准别怪我.
好学的话, 是STATA,
Stata最大的优势可能在于回归分析(它包含易于使用的回归分析特征工具),logistic回归(附加有解释logistic回归结果的程序,易用于有序和多元logistic回归)。Stata也有一系列很好的稳健方法,包括稳健回归,稳健标准误的回归,以及其他包含稳健标准误估计的命令。此外,在调查数据分析领域,Stata有着明显优势,能提供回归分析,logistic回归,泊松回归,概率回归等的调查数据分析。它的不足之处在于方差分析和传统的多变量方法(多变量方差分析,判别分析等)。
SAS有最强大的绘图工具,SAS的最优之处可能在于它的方差分析,混合模型分析和多变量分析,而它的劣势主要是有序和多元logistic回归(因为这些命令很难),以及稳健方法(它难以完成稳健回归和其他稳健方法)。尽管支持调查数据的分析,但与Stata比较仍然是相当有限的。
看看你有多少时间了, 如果时间很少很少, 那么,不妨用SPSS代替SAS,因为好学很多噢,上手快啦.而且方差分析也很好.但是局限性也很大,一般的统计可以应付, spss算业余水平吧,和sas比,好像industry也常用sas,
简单一点:混合模型分析,选择SAS;进行logistic回归则选择Stata;若是要进行方差分析,选择SPSS
或是,学个SPSS+Stata,时间上短,又能弥补互相的不足.
有时间再学SAS,matlab
不过,matlab主要是数学方面的数值计算,也有统计和金融工具箱,但是在金融领域不是常用.
总之总之,学这些超麻烦, 不知道说清楚没有.

根据你的需要选吧

作者: joysource 时间: 2007-12-6 15:10
Matrix Laboratory 满好用额
有个叫 Harald Uhlig的人,在1995写了篇文章,搞出个动态模型的优化算法
前人栽树后人乘凉,现在都是把他的代码直接拿来改了用的,网上有的LZ可以搜一下
http://www2.wiwi.hu-berlin.de/institute/wpol/html/toolkit.htm
作者: george_shi2001 时间: 2007-12-6 23:55
Yes,"For empirical research, one of Stata or SAS should be enough.".
But if you do pure econometrics, Matlab is necessary.
作者: galluplinc 时间: 2007-12-7 02:31
SPSS, SAS, RATS are all fine for business PhD, if you want to study forecasting & time series, you will definitely need to know EViews
作者: qiushi_wu 时间: 2007-12-7 09:22
提示: 作者被禁止或删除 内容自动屏蔽
作者: yession 时间: 2007-12-9 10:32
谢谢大家了 看样子支持SAS的人比较多 只是MATLAB的比较少
作者: lesliejy 时间: 2007-12-10 23:56
I am personally inclined to using Matlab and Mathematica for financial econometrics and computational finance (some recursive models). R is contributed by various people and it is completely free. therefore, you have to be cautious with R if you get some weired results. (Same for octave, which is a open-resource version of Matlab.)
Matlab works, for the moment, very well given my needs (OLS, GMM, SMM, DP, etc.) Most, if I can say so, matrix related operations can be efficiently evaluated by Matlab, as indicated by its name... To my experience, Mathematica can be used in solving (partial) differential equation, evaluating integrals, calculating derivatives. (I am not saying Mathematica can not do OLS, GMM, ...) Of course, when you need a very computational intensive code (with possible large memory requirement), you probably want to study c++, for instance.
作者: 立雪 时间: 2008-10-2 03:06
R is the most preferred language by some business professors. Harvard KSG requires all graduate students to learn R...and I believe their business school recommends that, too.
Then either one of SAS/STATA works, but in accounting SAS is preferred.
作者: kndx5 时间: 2008-10-2 03:21
well, R is prefered just because it is free.
作者: 立雪 时间: 2008-10-2 04:52
I don't think those professors care about which one is free and which is not....They don't need to pay for what they use.....
作者: himalaya1985 时间: 2008-10-6 04:28
R和matlab 不会太难的
作者: Dante 时间: 2008-10-8 05:25
sas for data manipulation
stata for regression
matlab for theoretical models
so u d better learn all of them.....
作者: qiushi_wu 时间: 2008-10-8 10:36
提示: 作者被禁止或删除 内容自动屏蔽
作者: xiaopanghu 时间: 2008-10-8 11:13
R就是速度太慢
欢迎光临 ChaseDream (https://forum.chasedream.com/) |
Powered by Discuz! X3.3 |