- Thread starter Dason
- Start date

Interesting. I always viewed you as a member of the R camp. Well , one member less. Could there have been, perhaps, irony in your statement? I honestly hope so... SAS is completely inadequate for implementing an original method or strategy, instead of relying on methods known for the last 20 - 70 years.

Hedge funds use Python, R, Matlab, C++ and C#. No waste there. Billions sucked out of clients. Apartments in downtown Manhattan... SAS is for people who completely depend on somebody else for statistical / econometric advice. When Trevor Hastie developed elastic nets 20 years ago, where did they appear first? No waste of software; rather the only software where the method was available for a while.

@noetsi - thought you may be interested as well.

https://support.sas.com/en/books.html?linkId=85529351

Did you get your PhD yesterday? Even people who got the PhDs in stats (that I know) in the early 2000's maybe up to 2015 even usually have a good working knowledge of SAS, just as you get more recent they have more and more R knowledge as well. The statisticians I know who were in school in the 60s-90s are almost ALL SAS and that was the gold standard for a long time. Your fanboy is showing.

Sure, SAS was the golden standard in the old days. Well-known. But statisticians who used SAS

Does anybody else have a guess of the current or 20-years ago best stats school in the US without looking online?

Does anybody else have a guess of the current or 20-years ago best stats school in the US without looking online?

What is there to guess? People who have taught me over the years:

Brad Efron - the inventor of bootstrap,

Jerry Friedman - the inventor of gradient boosting

Rob Tibshirani - the inventor of lasso,

Trevor Hastie - a co-inventor of elastic nets,

Thomas Cover - one of the leading names in Information Theory in the 20th century (RIP),

David Donoho - the most cited scientist in all the mathematical sciences (Mathematics, Statistics, Optimization, Computational Methods) in the US as of 2002,

Tze Lai - a leading expert in stochastic control.

Dear fellows, let us be civil to one another. We are here to help original posters, not to bicker. We are here to answer questions of those who make their first steps in Statistics. As much as our time and good will permits. Let us spend our energy on good things. Thanks.

Last edited:

Let us not fall to the level of foul language. Usually people use it when they have little else to say.... My PhD was received almost 20 years ago, from the #1 Statistics program in the US. When did you get your PhD? Was it in Statistics?

Sure, SAS was the golden standard in the old days. Well-known. But statisticians who used SAS*relied on somebody else* (SAS Institute programmers) to implement statistical methods for them. SAS was and is extremely clumsy for somebody who wants to implement his / her statistical or financial method from scratch. Unlike R, Python or Matlab. If you were to develop your own, highly customized hidden-factor model, if you were to develop your own, original method for estimating this model, there is no way you would do that in SAS.

Sure, SAS was the golden standard in the old days. Well-known. But statisticians who used SAS

I probably don't need a PhD in stats (nor do I have one, since you politely asked), to know that plenty of great statisticians used what was one of the most readily available and powerful programs at the time...

At the risk of more defensiveness, I'll leave it at that.

you can list the greatest teachers/mentors you want, but an appeal to authority doesn't make you any more credible because they did great things.

your post said, "**SAS is for people who completely depend on somebody else for statistical / econometric advice.**" This is not the same as ...

Politely disagree.** By your argument, a high school graduate from Uganda is likely to offer as accurate statistical assistance as a statistics PhD from Berkeley**... A solid multi-year training by leading experts goes a long way to becoming an expert in the field oneself (though it's not 100%). On your end, if you have not gone through formal statistical training, you may be missing parts of the big and small picture at times. Say, you may know classic epidemiology material but fail to see nuances in random fields or Bayesian nonparametrics, or sparse patter recognition... You might benefit from keeping an open mind and welcoming an opinion from those people who do not have such knowledge gaps.

Otherwise, let's move along.