Skip to main content
Announcements
Have questions about Qlik Connect? Join us live on April 10th, at 11 AM ET: SIGN UP NOW
cancel
Showing results for 
Search instead for 
Did you mean: 
_Lp
Contributor II
Contributor II

Random function

Hi,

I would like to generate random rows on shoe sales but when I use the random function it returns me a linear distribution. When I load my data, I will always get the same ranking order of shoe brands.

Below is my code:

LET vJOURS = '25' ;
LET vMOIS = '7' ;
LET vANNEES = '5' ;
LET vMODELE = '40' ;
LET vBOUTIQUE = '44' ;
LET vAGE = '45' ;
LET vPAYS = '5' ;
LET vVILLE = '2' ;
LET vCOULEUR = '3' ;
LET vTAILLE = '32' ;
LET vPRIX = '81' ;
LET vA = 0;

DO WHILE vA < 5

Ventes :
LOAD
ROUND(RAND()*'$(vJOURS)'+'$(vA)')+1 as JOURS,
ROUND(RAND()*'$(vMOIS)')+'$(vA)'+1 as MOIS,
FABS(ROUND((sqrt(-2*log(rand())) * cos(2* pi() * rand())* 2)+3)) as ID_ANNEES,
FABS(ROUND((sqrt(-2*log(rand())) * sin(2* pi() * rand())*+8)+43)) as ID_MODELES,
FABS(ROUND((sqrt(-2*log(rand())) * tan(2* pi() * rand())*10)+45)) as ID_BOUTIQUE,
FABS(ROUND(RAND()*'$(vA)')-2) as ID_SEXE,
ROUND(RAND()*'$(vTAILLE)'-'$(vA)')+'$(vA)'+1 as ID_TAILLE,
'1' as CNT
Autogenerate ROUND(RAND()*800000);

LET vA = vA+1;
LOOP

 

Do you have a solution to this problem so that I can get a real random distribution?

Thanks for your help,

Lucie

Labels (2)
6 Replies
Martijn_W
Contributor III
Contributor III

Hi Lucie

What should the results be like? At first glance it looks really, maybe overly, complicated.

_Lp
Contributor II
Contributor II
Author

Hi Marijn,

I would like to get a non-linear distribution of sales when I load data, which means to obtain a classification different from the previous one each time I load the data. But currently, I still find myself with the same ranking of the best sales of shoe brands, only the values(turnover, number of sales, ...) change. I would also like that by filtering on a country, the ranking differs from one country to another. However, in my case, the ranking of the best sales of shoes is always the same regardless of the country, which does not reflect the reality.

Thanks

Gabriel
Partner - Specialist III
Partner - Specialist III

@_Lp  try code below

SET vA =1;
DO WHILE vA < 5


Ventes :
LOAD
ROUND(RAND()*'$(vJOURS)'+'$(vA)')+1 AS JOURS,
ROUND(RAND()*'$(vMOIS)')+'$(vA)'+1 AS MOIS,
FABS(ROUND((SQRT(-2*LOG(RAND())) * COS(2* PI() * RAND())* 2)+3)) AS ID_ANNEES,
FABS(ROUND((SQRT(-2*LOG(RAND())) * SIN(2* PI() * RAND())*+8)+43)) AS ID_MODELES,
FABS(ROUND((SQRT(-2*LOG(RAND())) * TAN(2* PI() * RAND())*10)+45)) AS ID_BOUTIQUE,
FABS(ROUND(RAND()*'$(vA)')-2) AS ID_SEXE,
ROUND(RAND()*'$(vTAILLE)'-'$(vA)')+'$(vA)'+1 AS ID_TAILLE,
ROWNO() AS CNT
AUTOGENERATE ROUND(RAND()*800000);

LET vA = vA+1;
LOOP

_Lp
Contributor II
Contributor II
Author

Hi Gabriel,

I have tried this method before and it doesn't work.

Thanks

marcus_sommer

Here a few ideas how the generic values might be get more/broader distributed:

pow(rand() * sqrt(recno()), rand() * 7)

- Marcus

Martijn_W
Contributor III
Contributor III

I would use a way less complex fomula for randomness: round(rand()*83) for ID_MODELES.

You'd end up with a distribution like this:

distribution.png