Issue with Normal Distribution(Gaussian Distribution)

[manideep78]

Dear Community,

I had an issue with Normal Distribution Curve. My curve shows like below Image:

Here My Dimesnion is =if(BC_DEV<>' ', class(BC_DEV,0.01))

Expressions:

For Bars: =Count(DISTINCT  LENS_MASTER_SEQUENCE)

For Bell Curve: =if(BC_DEV<>'',NORMDIST(Class(BC_DEV,'0.01'),avg({1}TOTAL BC_DEV),Stdev({1}TOTAL BC_DEV),0))*2000

In the above expression I am Multiplying the expression with 2000 to get the curve in bell shape, otherwise it is like a flat straighline on X-Axis. And also I do not want to make the expression static by multiplying with 2000 or any other number.

If I make any selections I am not getting the exact curve for those particular filters. Please find the image below:

I want the Normal Curve to be calculated for any selections.

Please help me in achieving this. Thanks a ton in advance.

Regards

Manideep

[manideep78]

Can anyone please help me?

[jonathandienst]

Hi

Your chart is showing the fragment of the bell curve within your selection. You need your bell curve expression to take your selection into account. Try first with:

=if(BC_DEV<>'',NORMDIST(Class(BC_DEV,'0.01'),avg(TOTAL BC_DEV),Stdev(TOTAL BC_DEV),0))*2000

(remove the {1}'s)

HTH

Jonathan

[manideep78]

Yeah Jonathan Dienst. I am getting the curve. But I did not need the normal curve to be multiplied with 2000. Can you please tell me the best method. If I am not doing with 2000 or some other number I am getting a straight line along X-axis as I have my frequency in thousands.

[jonathandienst]

If you need to adjust the 2000 to reflect the selections you have made, so this will need to be calculated by an expression.

I can help you with details if you explain how you arrived at the value of 2000 for the full chart and post a sample of your qvw.

HTH

Jonathan

[manideep78]

Thanks Jonathan Dienst. Pls find my sample app attached. When I am using an expression instead of 2000 I am not getting a smooth bell curve as it was as earlier in this app. and If I made any selections I am not getting the ryt curve. and also I need 0.02 as my Class interval not 0.01. please find the attachment.

[jonathandienst]

I think that this is what you want. I have taken away the scaling and moved the curve to the right axis, and hidden the axis. Note that the curve will be distorted when only a few data points are selected as it is only calculated at these locations.

Jonathan

[manideep78]

I am unable to open your file. giving an error message like unable to load file.

[jonathandienst]

Looks like it did not upload correctly. Should be 2MB. Attached again...

[manideep78]

Thanks Jonathan Dienst.

Yeah it was similar to my requirement. But In the same App I need the Vertical Reference Lines, LSL, USL and also at 0. And also for some cases I want my curve to be starting at 0. But if you look the app with some filters like LOT = B00GZBS, B00GZBV  and B00GZBZ it is not starting at 0 level. Can you please look at this. I am attaching the images from the Sas report based on the above mentioned lots.. I want the same or something similar in Qlikview. Please help me in achieving this.