Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 
sparur
Specialist II
Specialist II

Standard Deviation Bell Chart

Hello colleagues

I need some help to build and understand how to get  nice "Bell curve" for my Answers data (please see at attachments).

For initial version I've used a recipe from Henric.


But the picture is not what I expected:

Capture1.PNG.png

I need to have something like this:

Capture2.PNG.png


Could someone explain me what should I do and where I was wrong in my current approach.


Might be I need to use another function (not the Normdist() )


Thanks a lot.

5 Replies
hic
Former Employee
Former Employee

I think you are doing the right thing. Normdist() is the function to use for a bell curve (a.k.a. Gaussian or Normal distribution).

The problem is that your Answer only has 6 distinct values. Hence, the Normdist() is calculated just for these six values - not for the values between. And so the curve will look a little crooked. It will look slightly better if you choose the "smooth" option though (Expressions - to the right of the "Line" check box).

HIC

sparur
Specialist II
Specialist II
Author

Hi Henric

Thanks for the explanation, with smooth it looks more kindly.

As you said, I can't get beautiful bell curve for my answers (1 to 6), so I think about Standard deviation instead of normal (Gaussian) distribution. Some more additional info I found here: Normal Distribution vs Standard deviation

Is it possible in QlikView? Do we have any specific function for it, like normdist() - for normal distribution?

Thanks,

Anatoly

hic
Former Employee
Former Employee

The standard deviation can be calculated using Stdev(). But it is not what you want here. You want a distribution of values with average and standard deviation as input parameters. Normdist() is probably the best choice.

HIC

Not applicable

Hello,

What is solution?

hic
Former Employee
Former Employee

  1. To use the Normdist() to create the bell curve. (If a bell curve is what you want.) What you should use as input (Mean and StDev) can be discussed though. Calculated from data? Or hard coded to something else?
  2. To use the "smooth" display option.

HIC

Normdist.png

Smooth.png