Scatterplot with no aggregation and with conditional expression

[Atish]

Hi Everyone,

I am trying to create a simple Scatter Plot without aggregation but with a condition.

My dataset is like this:

CD OFC_NO BKT VOL PCTG
XY 1 30 5 100
XY 2 30 15 100
XY 3 30 9 64.3
XY 3 100 5 35.7
AB 1 30 36 97.3
AB 2 100 1 2.7
AB 3 30 15 75
AB 3 45 1 5

The data is pre-aggregated.

I want to show a simple Scatter Plot (no aggregation) of OFC_NO as Dimension (so 1 bubble per OFC_NO), BKT as x-measure, PCTG as y-measure, VOL as size of bubble.

Following the examples given in Qlik community board, I see that I just have to give (BKT) instead of Sum(BKT) and (PCTG) instead of Sum(PCTG), and (VOL) instead of Sum(VOL).

However, an additional requirement is to restrict the chart to CD=="XY".

I can't do this using a filter panel/selection as I need to show the chart exactly for CD=="XY".

After looking up solved problems in the board, I see Qlik Sense lets me use a conditional expression like this but only if I do a Sum():

  Sum({<CD = {"FS"}>} BKT)

but this is not allowed:

  ({<CD = {"FS"}>} BKT)

Is there any way of doing this without aggregation?

Thanks,

Atish

[dplr-rn]

ignoring the tool used to display
whats the PCTG value for OFB_NO = 3 you want to display?

[Atish]

Hi Ranjith,

I would like the points to appear as in the graph I pasted (generated by pandas etc.).

So multiple values for same OFC_NO should be shown (it is guaranteed that CD + OFC_NO + BKT combination is unique).

In this case, OFC_NO = 3 would have two points, one for BKT=30 and one for BKT=100.

As in the graph I pasted, if multiple OFC_NOs have the same CD + BKT + PCTG combination, they would be shown as a single point to which they map.

This is for doing a quick trend analysis.

Thanks,

Atish

[Atish]

One more thing for the Qlik Sense Creators to know and possibly include in a future QS version for better Scatterplots:
By adding a small jitter, scatterplots can show multiple overlapping points e.g. Seaborn's stripplot can do this.:
https://stackoverflow.com/questions/8671808/matplotlib-avoiding-overlapping-datapoints-in-a-scatter-...

Of course, a small random jitter can be added to the data prior to calling a simple scatterplot generator (also discussed in above article).