Differential distribution from "cdf..." Scilab functions

Specific domains Statistics
1

What is the simplest way to get the differential normal distribution from Scilab “cdfnor”?

diff(P) from P=cdfnor(“PQ”,x,Mean(k)*ones(x),sqrt(Std2(k))*ones(x)); is clumsy, because I have to do some acrobatics with the x-values…

Heinz

2

Hello Heinz,

I had to deal with that too.
It is pitty that Scilab do only provide cdf functions

  1. you can “resample” the result from diff(P) with an interp1 like
    interp1(x(1:$-1),diff(P),x,"linear")
    but the “linear” option is sometimes creating NaNs
    You can try “spline”

  2. you can also, if your vector is large say,
    y=diff(P); y($+1)=y($)
    This is copying the last value of your derivative vector twice
    Maybe you can live with this trick. I do … sometimes.

  3. you can install STIXBOX, DISTFUN will come with it
    In this DISTFUN tool box there is a function called distfun_normpdf that will do what you want.
    y=distfun_normpdf(x,mu,sigma)

My hope is, one day, all the Statistics functions in STIXBOX and CASCI will be bundled in SCILAB (No need of Toolboxes)

BR
Fanch

3

Hello,

In the above case the mentioned toolboxes are made available along Scilab’s versions. So please let’s be constructive, letting Scilab become bigger and bigger is not a solution to particular needs.

S.

4

Hi Stephane,

I’m ok to leave the Statistic functions outside in a Toolbox. (What is existing inside Scilab, STIXBOX, CASCI … all in one, no duplicates, no conflicts). This looks like the same dream as this page: The-Ideal-Statistics-Module

Then we would need someone to maintain it. I would love to be capable of doing it (but time and competencies …)

I already discussed this at length with Vincent & Antoine.

Recently, I found a mistake in STIXBOX, reached out to the owners thru Gitlab but no answers. Fortunately, Vincent fixed it and we should have a new version soon. So that is why I was thinking, maybe, if it was all bundle in Scilab there will be someone to take care of it.

A good discussion topic for sure

BR
F