Logarithmic Histogram

[MoeSzyslak]

All --

I am attempting to fit a lognormal distribution to a histogram with logarithmically-spaced bins, but am not having much luck. I suspect that at least part of my problem is the old "linear fit, logarithmic plot" problem, but I can't seem to shake it this time. Any ideas?

As always, thanks in advance....

Matt

[ptc-1368288]

On 4/9/2010 12:33:33 AM, Moe_Szyslak wrote:
>All --
>
>I am attempting to fit a
>lognormal distribution to a
>histogram with
>logarithmically-spaced bins,
>but am not having much luck. I
>suspect that at least part of
>my problem is the old "linear
>fit, logarithmic plot"
>problem, but I can't seem to
>shake it this time. Any ideas?
>
>As always, thanks in
>advance....
>
>Matt
_____________________________

Your Z vector is now raw data, too many 0 values or near zero. If it would be from collection, then the ratio of the max to the min indicates a very poor collection. I don't understand your fabrication. You must test the data first and as you can see the histogram is not lognorm.

You could probably fit several distributions to that histogram

jmG

[RichardJ]

It doesn't matter what the x-spacing is. Minerr just minimizes the y-residuals.

Dividing y in the solve block by a negative number does not help 😉

Richard

[ptc-1368288]

It could be that the original data are paired.
It would enable a weighted histogram "Paul style".
w/o the raw data = end of project.


jmG

[ptc-1368288]

On 4/9/2010 7:43:49 PM, jmG wrote:
>It could be that the original
>data are paired.
>It would enable a weighted
>histogram "Paul style".
>w/o the raw data = end of
>project.
>jmG
______________________________

An extraordinary good Gaussian fit,
that puts me out of a context w/o context.

jmG

[MoeSzyslak]

Richard, Jean --

Thanks so much for the help -- sometimes I can be a little dense.

Matt