Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 
michael_klix
Creator II
Creator II

heatmap with continuous x-axis

Hi everybody,

I am looking for a heatmap solution that allows me to show the value for about 20 different measurements over time.

In principle, the solution is a pivot table:

The name of the measurement should be displayed vertically and the time horizontally. The data itself then defines the colour of the respective cell using the colormix wizard (for text and background to erase the numbers in the cells)

Easy thing...BUT:

My time dimension has about 1000 or even more variants which ends up in >1000 columns.

Is there a way to make the columns infinitely small? In my QV11.2 it seems there is a minimum column width of about 1 character or so. But I want the whole range of time to be displayed on one screen only.

Any ideas?

All solutions I found online work with just a few columns on the x axis.

Keep Qliking

Michael

1 Solution

Accepted Solutions
michael_klix
Creator II
Creator II
Author

Problem solved, surprisingly easy:

Remember, there is a time series of about 10.000 points in time, each has a value fro a set of measurements.

What I used was a stacked barchart with the expression

=only(Time)-Above(Time)

what it does is simply taking the first timepoint as a baseline and adding the time increment for each point in time. Given that time starts with zero (did I mention?) the stacked bar in the end is exactly the full period of time.

This bar now exists for each measurement and the VALUE of the measurement itself is only reflected in the COLOR of the bar. That's what makes it a heatmap.

That's how it looks:

Thanks for all the ideas however

Keep Qliking

Michael

View solution in original post

4 Replies
marcus_sommer

Hi Michael,

could it be a solution to cluster the time-values with something like: time(class(YourTime, 1/24/60 * 5), 'mm') ?

- Marcus

michael_klix
Creator II
Creator II
Author

Hi Marcus,

unfortunately not really as we need the individual timestamps to be selectable.

My workaround was a faked chart with continuous time on the x-axis and a stacked bar chart with measurement type as 2nd dimension. To get a unique height for each measurement, the expression is simply “=1” So each stacked bar is 20 units high.

Disadvantage: The y-axis gets the values 1-20 instead of the names of the measurement and the legend does not synchronize with the levels of each measurement in the stacked bars…

Keep qliking

Michael

Von: Marcus Sommer

Gesendet: Montag, 11. Juli 2016 16:28

An: Klix, Michael

Betreff: Re: – heatmap with continuous x-axis

Qlik Community <https://community.qlik.com/?et=watches.email.thread>

heatmap with continuous x-axis

Antwort von Marcus Sommer<https://community.qlik.com/people/marcus_sommer?et=watches.email.thread> in QlikView App Development - Komplette Diskussion anzeigen<https://community.qlik.com/message/1080593?et=watches.email.thread#1080593>

marcus_sommer

Hi Michael,

maybe it's better to separate the pivot-heatmap from legend-informations and selectable timestamps. I think I would do some cluster on the time-level and maybe add some informations within the expression, something like:

= 1 & chr(10) & chr(10) & concat(Timestamp, chr(10)) & chr(10) & 'some further details'

and use a listbox or a straight-table beside the pivot with the single timestamps and maybe a count of them and other descriptive informations.

- Marcus

michael_klix
Creator II
Creator II
Author

Problem solved, surprisingly easy:

Remember, there is a time series of about 10.000 points in time, each has a value fro a set of measurements.

What I used was a stacked barchart with the expression

=only(Time)-Above(Time)

what it does is simply taking the first timepoint as a baseline and adding the time increment for each point in time. Given that time starts with zero (did I mention?) the stacked bar in the end is exactly the full period of time.

This bar now exists for each measurement and the VALUE of the measurement itself is only reflected in the COLOR of the bar. That's what makes it a heatmap.

That's how it looks:

Thanks for all the ideas however

Keep Qliking

Michael