
1. Re: Calculation  Normal Distribution
Alex Kerin Apr 23, 2013 5:38 AM (in response to buildstar)There is not one. You would have create one. There are some threads one normal dist: https://www.google.com/webhp?sourceid=chromeinstant&ion=1&ie=UTF8#safe=off&output=search&sclient=psyab&q=tableau%20normal%20distribution&oq=&gs_l=&pbx=1&fp=7999b0078f7a181e&ion=1&bav=on.2,or.r_cp.r_qf.&bvm=bv.45512109,d.dmQ&biw=1920&bih=989

2. Re: Calculation  Normal Distribution
buildstar Apr 23, 2013 6:15 AM (in response to Alex Kerin)Yes I tried but running into difficulties framing a calculation.
Can you please help?
Found this thread too:
http://community.tableau.com/thread/118995
Thanks

3. Re: Calculation  Normal Distribution
Brad Llewellyn Apr 23, 2013 8:53 AM (in response to buildstar)Gallop,
I previously made a blog post about doing something like this
http://breakingbi.blogspot.com/2013/03/conducting2sampleztestintableau.html
With your idea and a renewed passion. I was able to create the attached workbook.
Let me know if this helps,
Brad Llewellyn
Associate Consultant
Mariner, LLC
brad.llewellyn@marinerusa.com

Book3.twbx.zip 98.3 KB


4. Re: Calculation  Normal Distribution
Alex Kerin Apr 23, 2013 9:08 AM (in response to Brad Llewellyn)Nice work Brad  your contribution to the forums has been phenomenal. Have you added any of these to the tableau calculation reference library?

5. Re: Calculation  Normal Distribution
Brad Llewellyn Apr 23, 2013 9:13 AM (in response to Alex Kerin)Alex,
If one were so inclined, where would one find such a magical place?

6. Re: Calculation  Normal Distribution
buildstar Apr 23, 2013 10:15 AM (in response to Brad Llewellyn)Brad,
Excellent solution.Thanks for your help.
However I have the zscore formula to be:
((Calc)LOOKUP([Calc], 1))/POWER((((([Calc])*(1[Calc]))))/SUM([matched])+(((LOOKUP([Calc], 1))*(1  LOOKUP([Calc], 1)))/(LOOKUP(SUM([matched]), 1 ))),0.5)
And I use the following formulae for p value
IF INDEX() <> 3 AND INDEX() <> 1
THEN
IF ABS([ZScoreCTR] ) <= 2.2 THEN ( .5  ( ABS( [ZScoreCTR] ) * (4.4  ABS( [ZScoreCTR] ) ) / 10 ) ) * 2
ELSEIF ABS( [ZScoreCTR] ) <= 2.6 THEN .02
ELSE 0
END END
I have value populated for the zscore but yet the value for pvalue is shown as 0.00% . Trying to recify it but though could use an expert help to find reason I am unable to see the value for pvalue
Thanks

Untitled.png 74.2 KB


7. Re: Calculation  Normal Distribution
Brad Llewellyn Apr 23, 2013 10:09 AM (in response to buildstar)Gallop,
Could you post your .twbx or a screenshot?
Thanks,
Brad Llewellyn
Associate Consultant
Mariner, LLC
brad.llewellyn@marinerusa.com

8. Re: Calculation  Normal Distribution
buildstar Apr 23, 2013 10:16 AM (in response to Brad Llewellyn)Brad,
Attached the screen in same message.
Thanks

9. Re: Calculation  Normal Distribution
Brad Llewellyn Apr 23, 2013 10:25 AM (in response to buildstar)Gallop,
Look back at your PValue formula, you are telling it to return zero for ZScores more extreme than 2.6. Also, mathematically, any ZScore more extreme than 3 has such as an astronomically small probability, it is commonly called zero. This is all as it should be. Your report looks great!
Thanks,
Brad Llewellyn
Associate Consultant
Mariner, LLC
brad.llewellyn@marinerusa.com

10. Re: Calculation  Normal Distribution
buildstar Apr 23, 2013 11:40 AM (in response to Brad Llewellyn)Brad,
Thanks a lot for your excellent response.
I have attached a screenshot from excel. Still how do I arrive at pvalue based on the formula attached previously ?
Thanks

Untitled.png 6.2 KB


11. Re: Calculation  Normal Distribution
Brad Llewellyn Apr 23, 2013 11:52 AM (in response to buildstar)Gallop,
Excuse me for getting a little technical here.
A PValue is
2 * Prob( X > abs( [ZScore] ) )
In layman's terms, this means "The theoretical probability of observing an outcome at least as extreme as the one actually observed."
What Excel is calculating is
Normal CDF = Prob( X < [ZScore] )
Therefore, the formula you were using in Excel is incorrect. Your Excel formula should be
= 2 * (1  NORM.DIST(ABS(G12),0,1,TRUE) )
I apologize for not recognizing this earlier.
Hope this helps,
Brad Llewellyn
Associate Consultant
Mariner, LLC
brad.llewellyn@marinerusa.com

12. Re: Calculation  Normal Distribution
Alex Kerin Apr 23, 2013 12:36 PM (in response to Brad Llewellyn)http://community.tableau.com/community/calculationreferencelibrary
If you go to ideas, it is then one of the listed headings on the right. Why it isn't linked from all forum sections I don't know (Dustin Smith )

13. Re: Calculation  Normal Distribution
buildstar Apr 24, 2013 12:44 AM (in response to Brad Llewellyn)Excellent solution....Thanks a ton Brad...awesome calculation !!!

14. Re: Calculation  Normal Distribution
Sriram Natesan Feb 26, 2016 4:22 AM (in response to Brad Llewellyn)Hi Brad,
Its an interesting solution. I have gone through the other blog also. I would need a small clarification in the calculation where you have used 2.2 in the IF construct and finally divided by 10.
I would like to know whats the baseline behind this formula you have used.
I came across another blog where the CDF is calculated. That also seems interesting just pasting the link here
Cumulative Distribution Function (Normal Distribution)
IF INDEX() <> 3 AND INDEX() <> 1
THEN
IF ABS([ZScoreCTR] ) <= 2.2 THEN ( .5  ( ABS( [ZScoreCTR] ) * (4.4  ABS( [ZScoreCTR] ) ) / 10 ) ) * 2
ELSEIF ABS( [ZScoreCTR] ) <= 2.6 THEN .02
ELSE 0
END END