Add CDFs to Graph Builder

[BHarris]

Recommend adding the ability to create CDF plots within Graph Builder.  

Add a new "Connection" type -- maybe "Extended Step", that would draw the first step from the left edge starting at 0, and from the last point to the right edge to represent a true CDF.  Then, document the approach, and consider adding these settings as some type of preset.

[hogi]

could you please also add the

Maybe add a setting which allows the user to choose between 
- max (like now)

- midpoint (like Prob Scores in Distribution)

- individual points (like Cumulative Probability and standard Col Rank)

- min (if somebody needs this as well)

[TorP]

So, seems like everybody is satisfied :-).

Still, I think there could come more out of this discussion, but possibly this may be the wrong wish for what I have in mind?

If so, just let me know.

What I’d propose again is to add Distribution Fitting w/ Confidence Intervals to Graph Builder, similar to the “Life Distribution” module.

(And sorry that the post may gonna become a bit long, but I’d like to show the “reasoning”.)

Available in JMP is:

Big Class.jmp

CDF in Graph Builder possible but only w/ lines, not data points

Adding a column w/ 1s to Big class.jmp

Now CDF in Graph Builder w/ points is possible

Using Smoother even a “Guide to the eye” fitted line can be added in Graph Builder.

A real “Line of Fit” for a distribution is not possible (or at least I did not manage :-)).

And moreover the “Confidence of Fit” can be added

It is also possible to use Probability Scales like Normal or Weibull instead of Linear.

(Works only w/ lines or points, though, Smoother lines and Confidence shapes are removed when changing the Y-axis scale.)

So, the actual graphical ingredients for a “Life Distribution”-like plot appear to be already available in Graph Builder (of course I don’t know what’s actually going on “under the hood”).

What I’d propose is to enhance the “Smoother” or better the “Line of Fit” option to fit actual distributions to the data points.

Could be under “Fit” and then a selection of the actual distribution.

And an option to add and set the Confidence Interval.

Imho, this is not just nice to have in Graph Builder, but a real enhancement to just plotting CDFs.

Since you see directly from the distribution fit (which extrapolates the actual data) what lifetimes or failure rates you may achieve.

This feature is often used in quality departments but likely less so for std. electrical data analysis at 0h.

I’m particularly interested to see this in Graph Builder, since I would like to get rid of Minitab (where this is easily done), but I could imagine other people could like it as well.

And if someone really wants to dive deeper, there’s always the option of the “Life Distribution” module, which gives much more data and options then the proposed Graph Builder implementation. But unfortunately the “Life Distr.” has not the split view options like Graph Builder.

Still, even the “Life Distribution” module could use some improvement, since besides its impressive graphics, it lacks a legend and info about the used probability and confidence interval (see wish Life Distribution - one Page Overview - JMP User Community).

In my opinion, adding distribution fitting to Graph Builder would be even more useful than adding a split view for "Life Distribution".

[hogi]

Agree.

really a pity that Minitab and Life Distribution can provide the fit - and Graph Builder can't *)

*) actually, it seems - GraphBuilder can - somehow -  but just for the case of the manually calculated Cumulative Probability - and with a quite narrow confidence interval ...
https://community.jmp.com/t5/Discussions/how-to-make-cumulative-probability-plots-in-JMP/m-p/806944/...