topic Re: What is the best programming: New dimension in table or using arrays in SAS Programming

What is the best programming: New dimension in table or using arrays

andraro — Fri, 03 May 2019 18:32:20 GMT

Hi All,

I am new to SAS but I have been working with databases for a long time. I am struggling to understand what is the best way to work with SAS. I have variables that are dependent on time and from my understanding of database structuring is that there should be a column with the time step. Although this means that all the characteristics of the item will be duplicated many times which creates a large database?

As in SAS there are Arrays (which are just a grouping of variables) this makes the database more compact but what about the runtime is SAS not more efficient in doing line by line instead of a do loop?

I hope that my explanation is clear and that someone can help me in my dilemma.

Thanks to all for helping out.

Re: What is the best programming: New dimension in table or using arrays

PGStats — Fri, 03 May 2019 19:25:41 GMT

A good tradeoff is to use two tables. One for item characteristics, the other for the time series, with an itemID key to link the two tables.

Going wide (arrays) will prove cumbersome in all aspects of data management and analysis.

Re: What is the best programming: New dimension in table or using arrays

andraro — Fri, 03 May 2019 19:30:37 GMT

Indeed one of the disadvantages of using arrays is the data handling.

How do you handle the fact that you need the characteristics to do the calculation of the time series.

Re: What is the best programming: New dimension in table or using arrays

PGStats — Fri, 03 May 2019 20:56:41 GMT

When required by a given procedure: create views that join item characteristics to time series. A view generates the expanded dataset only when it is read, the whole data never has to be stored anywhere.