Parallel Coordinates/Slope/Ranking Graph

[#1]

Hi everyone,

I need help building a chart.  I have data that scores Markets by different measures, e.g.:

Market	Size Score	Growth Score	Attractiveness
A	3.0	4.0	7.0
B	2.0	6.0	4.0
C	5.0	1.0	3.0

I want a line chart where each market is a line and Size, Growth, and Attractiveness appears on the X axis. The values would appear in the Y axis as they do in the above table.

The chart would look like this:

The only solution I know of to build this chart is to use the calculated dimension/expressions in the Parallel Coordinates chart provided in the QlikTech demo:

However, this chart is more complicated than my concept is. I do not need to calculate frequency, ranking, %s, or anything - I just want to show the data that I have. I tried to use a simpler solution, but it relies on using one time dimension - Year. I want to compare several dimensions.

Does anyone know a potential solution? i thought the best would be to manipulate the calculated dimension/expression of the parallel coordinates chart - could anyone help me do this? The two main formulas are:

Dimension:

=$(='if('&concat('NOT IsNull(['&CustomerMix&'])',' AND ')&', hash128('& concat('['&CustomerMix&']',',')&'))' )

Expression:

dual(

$(=concat('if(CustomerMix='&chr(39)&CustomerMix&chr(39)&',concat(distinct ['&CustomerMix&']))','&'))&'

'&num( 100*node_percent, '#,##0.#', '.' , ',' )&'%'

,

$(=concat('if(textcount(total ['&CustomerMix&'])=0,if(CustomerMix='&chr(39)&CustomerMix&chr(39)&',0.1+((min(['&CustomerMix&'])-min(total ['&CustomerMix&']))/(max(total ['&CustomerMix&'])-min(total ['&CustomerMix&']))),0),if(CustomerMix='&chr(39)&CustomerMix&chr(39)&',0.1+((aggr(nodistinct rank(['&CustomerMix&'],4),['&CustomerMix&'])-1)/rangemax( (count(distinct total ['&CustomerMix&'])-1) ,1 )),0))','+'))

)

Thanks,

Andrew

[Anonymous · Accepted solution]

i may be missing something or may be you are overthinking this...please see attached

[#3]

Ohhh. You just have to gather the different score types into one dimension!

That is extra work, but I'm so happy I can build the chart now. Thank you jsaradhi!

[Anonymous]

you are welcome...if you are getting this data from a database, try and move the crosstable part into the database using relevant transpose functions. i do this as crosstable function takes up too much memory and sometimes is not clean!

[#5]

In this case, the data comes from a more simple Excel file. However, I cannot change it. Is there a way to transpose those dimensions into one column with a calculated chart dimension? Otherwise, I was planning to construct a new table in a second Excel tab.

[Anonymous]

if the data size is not too big, you may be ok with doing it in Qlikview. my above post is more of a fyi to keep in mind. i would cross table it once instead of calculating the dimensions in UI qvw each time a user makes a selection.

[#7]

Thanks for the warning, but if I wanted to do it in the QVW, how would I do it?

[Anonymous]

take a look at the script in the qvw i attached for an example.

[#9]

Ah, I understand now. Thank you!

[Anonymous]

np