How to get started in learning IML?

[novinosrin]

Hello everybody, I am pretty keen on learning IML and am wondering where and how to start.

i. I would like your opinion on the prerequisities before learning IML- Do I need very strong linear algebra, probability and calculus knowledge and strong understanding?

ii. Does SAS documentation cover from the very basic to the advanced topics at length including the necessary concepts pertaining to real world applications?

iii. Can you mention any industry applications where IML is "widely" used as apposed to  SAS STAT and of course the foundation base SAS.

iv. I have access to a book by author Rick in safarionline library, is that targeted more towards highly experienced users or an IML beginner can benefit too?

Thank you & Regards,

Naveen Srinivasan

[Reeza]

I can't answer all your questions.

1. I would expect you should know some basic linear algebra to start out or it won't make sense. Concepts such as a matrix, array, row/column operations, matrix math, invert/transpose matrices are basics that you should understand theoretically. 

2. SAS documentations will cover the applied portion as well as how to use the functions/modules/features. It will not necessarily teach 'real world applications'. This would be a big ask, IMO.

3. IML is more similar to R/Python, and I can't point to a specific area where it may be more predominant. It's typically used when you're likely to be developing your own algorithms or models, not working with standard models. Another use is for faster processing, sometimes things are faster in IML  - and no, I can't come up with a specific example at the top of my head, but if you search on here you'll probably find a few examples.

4. No idea.

Other resources:

Lexjansen.com -> search intro to IML for user written papers

SAS docs are pretty good in coverage

SAS IML Studio -> ? Not sure but worth checking out, IMO

Rick Wicklin's blog 

He'll probably answer at some point anyways so he can comment on the suitability of the books 🙂

[Rick_SAS · Accepted solution]

How to get started in learning IML?

[ New ]

> Hello everybody, I am pretty keen on learning IML and am wondering where and how to start.

I recommend you start with the resources in the article "Ten tips for learning the SAS/IML language"  I also recommend that you browse and/or subscribe to The DO Loop blog: blogs.sas.com/content/iml

i. I would like your opinion on the prerequisities before learning IML- Do I need very strong linear algebra, probability and calculus knowledge and strong understanding?

You do not need those prerequisites to start learning. However, the more you know, the more you can do! Areas that people use SAS/IML for include statistical data analysis (implementing new statistical tests), multivariate data analysis (requires matrix computations), nonlinear optimization (calculus helpful), numerical analysis, simulation, and bootstrap methods.

ii. Does SAS documentation cover from the very basic to the advanced topics at length including the necessary concepts pertaining to real world applications?

The documentation focuses on the language and the syntax.  Most of it is basic. Several chapters ("General Examples", "Robust Regression", Nonlinear Optimization") include applications, but the emphasis is always learning how to use the functions. For advanced topics, see SAS Global Forum papers, The DO Loop,  and chapters of Statistical Programming Using SAS/IML Software (Wicklin , 2010)

iii. Can you mention any industry applications where IML is "widely" used as apposed to  SAS STAT and of course the foundation base SAS.

Finance, insurance, medicine, and pharmaceuticals are three industries where SAS/IML is used widely. You can do a Goolge search to find partiicular applications within those areas.

iv. I have access to a book by author Rick in safarionline library, is that targeted more towards highly experienced users or an IML beginner can benefit too?

The book  Statistical Programming Using SAS/IML Software (Wicklin , 2010) starts out at the beginning and is appropriate for beginners. The book Simulating Data with SAS (Wicklin, 2013) assumes some familiarity with SAS/IML, but tries to explain the relevant ideas as necessary.

[Rick_SAS · Accepted solution]

i. I would like your opinion on the prerequisities before learning IML- Do I need very strong linear algebra, probability and calculus knowledge and strong understanding?

You do not need those prerequisites to start learning. However, the more you know, the more you can do! Areas that people use SAS/IML for include statistical data analysis (implementing new statistical tests), multivariate data analysis (requires matrix computations), nonlinear optimization (calculus helpful), numerical analysis, simulation, and bootstrap methods.

ii. Does SAS documentation cover from the very basic to the advanced topics at length including the necessary concepts pertaining to real world applications?

The documentation focuses on the language and the syntax.  Most of it is basic. Several chapters ("General Examples", "Robust Regression", Nonlinear Optimization") include applications, but the emphasis is always learning how to use the functions. For advanced topics, see SAS Global Forum papers, The DO Loop,  and chapters of Statistical Programming Using SAS/IML Software (Wicklin , 2010)

iii. Can you mention any industry applications where IML is "widely" used as apposed to  SAS STAT and of course the foundation base SAS.

Finance, insurance, medicine, and pharmaceuticals are three industries where SAS/IML is used widely. You can do a Goolge search to find partiicular applications within those areas.

iv. I have access to a book by author Rick in safarionline library, is that targeted more towards highly experienced users or an IML beginner can benefit too?

The book  Statistical Programming Using SAS/IML Software (Wicklin , 2010) starts out at the beginning and is appropriate for beginners. The book Simulating Data with SAS (Wicklin, 2013) assumes some familiarity with SAS/IML, but tries to explain the relevant ideas as necessary.

[novinosrin]

@Rick_SAS Very neat and thank you. I really appreciate the response.