Business Intelligence

If you need to export SAS data into a XML file than you have several options.
One possibility is a manual export via datastep.
Please take into account that the XML „child“ data must be stored horizontal in the SAS table.
You can access this SAS „child“ data later by using n-dimensional arrays.

The following example shows how it works.
The „+(-1)“ is a pointer to avoid blanks in the final XML file.

filename outxml „[yourPath]“;
data _null_;
 file outxml;
 set input NOBS=Lst;
by custid;

If you want to import all files of a folder with a specific name or type you can use wildcards to store them in a SAS dataset.
You can use this example:

/*
this datastep will import all
ASC files of the specified folder
*/
data WORK.share;
infile ‚D:\yourPath\*.asc‘
delimiter=’09’x
MISSOVER
DSD
lrecl=32767 ;
format date mmddyy10. ;
informat date mmddyy10. ;
input
date
;
run;

If you have problems using the autoexec.sas in combination of SAS Enterpise Guide you can use my instructions:

1. run Enterprise Guide
2. Click on Tools
3. Click on Enterprise Guide Explorer
4. In the Folder „Servers“ right click on the SAS Server of your choice
5. If possible click on „Disconnect“ otherwise click on „Properties“
6. Click on Options then you can type in a „SAS startup statement“
7. Here you type in the startup instruction for your autoexec.sas.
For Example:
%include ‚/path/autoexec.sas‘;
Don’t forget the semicolon!

*
SAMPLE SIZE CALCULATION
How many customer should we randomly select to get significant results?
– Customer Base: 10.000 customer
– Standard Deviation calculated by PROC MEANS
– Probability value: 5%
– Margin of Error: 1 Euro

Detailed Documentation:
 http://www.isixsigma.com/library/content/c000709a.asp

Simulation of Customer Base Table:
 acct_num = Accountnumber
 fee = monthly fee in Euro per customer
;

data customer_base(drop=i);
do i=1 to 10000 by 1;
 acct_num=i;
 fee= 100 + round(25*ranuni(1),1);
 output;
end;
run;

proc means data=customer_base;
 var fee;
 title ‚Descriptive Statistics using Proc Means‘;
 output out=customer_statistics;
run;

*store Standard Deviation in macro variable for later calculations: ;
proc sql noprint;
 select fee into :std
 from customer_statistics
 where _stat_=‚STD‘
;
quit;

*
Calculation of the sample size:
sample_size = (critical value Z alpha/2 * Standard Deviation of the Population) / margin of the Error)**2
Z alpha/2 is calculated with the Table of Standard Normal Distribution
;
data sample_size;
 sample_size=round(((1.96 * &std)/1)**2,1);
run;
* In this example a Minimum of 200 customers is needed to get significant results;