Structuring Wide Data

Fri, 12 Jun 2015 20:43:32 GMT

I have a really wide market research dataset. Each respondents is a row and demos / question answers are the columns. Which makes a very wide dataset - 6000 columns. Data like so:

Respondent	Age	Gender	Q1	Q2	. . .	Q6000
1	25-30	m	1	2	. . .	1
2	30-35	m	2	3	. . .	2
3	25-30	m	1	3	. . .	5
…	…	…	…	…	. . .	…
4000	25-30	f	4	5	. . .	4

From this I need to create a table something like so:

Age	Gender	Respondents	Respondents meeting criteria
25-30	m	800	100
25-30	f	1200	50
30-35	m	1500	300
30-35	f	500	72

The dimensions for this table can be up to 10 of any of the 6000 columns chosen dynamically.

Using that table as 6000 columns by 4000 rows, I can create the app it is however a bit slow.

I am thinking that if I transpose the data as such - then I can improve performance. In this type of table I have no problem with one dimension, however, adding dimensions to get the result table above is escaping me.

My other thought is to break up the 6000 column table into logical use groups to enhance performance.

Any thoughts?

Thanks

1	Age	25-30
1	Gender	m
1	Q1	1
1	Q2	2
1	. . .	. . .
1	Q6000	1
2	Age	30-35
2	Gender	m
2	Q1	2
2	Q2	3
2	. . .	. . .
2	Q6000	2
3	Age	25-30
3	Gender	m
3	Q1	1
3	Q2	3
3	. . .	. . .
3	Q6000	5
. . .	. . .	. . .
4000	Age	25-30
4000	Gender	f
4000	Q1	4
4000	Q2	5
4000	. . .	. . .
4000	Q6000	4

Re: Structuring Wide Data

MarcoWedel — Fri, 12 Jun 2015 23:40:18 GMT

you could use a CrossTable load to get the data model you are looking for.

regards

Marco

topic Re: Structuring Wide Data in QlikView

Structuring Wide Data

Re: Structuring Wide Data