| Author |
 Topic  |
|
|
davis_wolf
United Kingdom
2 Posts |
 Posted - 08/20/2014 : 3:08:33 PM
|
Origin Ver. and Service Release (Select Help-->About Origin):
Operating System:Origin 9.0
Guys can you help me with the following problem in fitting of LogNormal function? I'm an amateur and totally crazy about this right now...
I need to fit a histogram with LogNormal function showing on WIKI like this:
but the function Origin offers me is like this:
I tried to set y0=0 and A=1, but the fitting result shows the standard deviation w is way bigger than real situation....it's not right....it's like this:
Can body teach me how to fit the histogram right with the Lognormal function I want?
MANY THANKS!!! |
Edited by - davis_wolf on 08/20/2014 3:09:27 PM |
|
|
Shirley_GZ
China
Posts |
Posted - 08/20/2014 : 10:11:48 PM
|
Hi Davis,
In the upcoming Origin version, there is a tool Distribution Fit which can be used to perform distribution fit. It can output the LogNormal as a distribution directly.
Currently, you can plot a histogram, then double-click it to open the Plot Details dialog. Go to the Data tab, select LogNormal for the Type in the Distribution Curve part. If you want to see the bin data, you can check the Add Distribution Curves check box, and then click Go button to go to the Bin Worksheet. In the Bin Worksheet, the parameters are in the Comment cell of col(Lognormal).
Originlab Technical Service Team |
Edited by - Shirley_GZ on 08/20/2014 10:21:40 PM |
 |
|
|
davis_wolf
United Kingdom
2 Posts |
Posted - 08/21/2014 : 09:59:37 AM
|
quote:
Originally posted by Shirley_GZ
Hi Davis,
In the upcoming Origin version, there is a tool Distribution Fit which can be used to perform distribution fit. It can output the LogNormal as a distribution directly.
Currently, you can plot a histogram, then double-click it to open the Plot Details dialog. Go to the Data tab, select LogNormal for the Type in the Distribution Curve part. If you want to see the bin data, you can check the Add Distribution Curves check box....
Thank you Shirley! That really helps a lot and now I've got the histogram with proper fitting curve!
But another problem is that now the mean value seems right, but the Standard Deviation doesn't...you might need to take a look at the picture below, the ln SD=1.027348670751, which means SD=2.7936. But the data range on X axis is just 0 to 0.015, how can the Standard Deviation be so big?
Sincerely looking forward to your help
Sincerely looking forward to your help |
|
|
|
Shirley_GZ
China
Posts |
Posted - 08/26/2014 : 03:30:26 AM
|
Hi,
1. Actually, the ln SD is w, no need calculate exp(ln SD) again.
2. About why the SD is so big, please see below example,
I assume that the minimum is 1E-5, ln(1e-5)= -11.51292546497, and maximum is 0.015, ln(0.015)= -4.1997050778799. ln(0.015)- ln(1e-5)= 7.31322. So the ln SD=1.027348670751 is possible.
3. ln SD is just a label, not the ln value of SD. So SD shouldn't be calculated by the exp(ln SD) . If you want to calculate it, please use the below equation.
SD=exp(a+0.5*b^2)*sqrt(exp(b^2)-1)) where a=ln(xc), b=ω
Originlab Technical Service Team |
Edited by - Shirley_GZ on 07/08/2015 12:07:41 AM |
|
|
| |
Topic |
|
|
|
Need help on LogNormal function fitting