It's an implementation of a new statistical programming language, now up and running in beta mode, that produces statistical tables really fast , with a really small effort in coding.

Since the software is in beta mode, I've already run some examples:

Typical AE table (N % p-value(chisquare) , each body-system, patients who do not have (each bodysys), and of course body-system*preferred-term) ->

That's 12 lines of code, baby! (SAS, you need around 600!)

Typical baseline characteristic ( but no continuous ), 3 categorical variables going down, one category across(treatment group) , N % p-value(chisq) ->

That's 9 lines of code , man! (SAS, hard pressed to get it under 400)

Linear model example: pairwise (t-statistic and p-value, ends up being 4x4, in the leftmost section of paper), model coefficients( t-statistic and p-value, one column 4x1 ), and F-statistic (with p-value, in the rightmost part of page) ->

That's 11 lines of code , man (SAS, hard to say, probably 400 lines of code, or more)

This is not a macro library. This makes use of tree data structures in a much more sophisticated way than SPSS/SAS/SQL.

The current version is on Linux and outsources(gives work to do) to vilno data transformation and R. This DOES NOT mean that the product forces the end user to look at vilno code or R code. A version that outsources to SAS/BASE and SAS/STAT is not yet in beta mode

## No comments:

## Post a Comment