| T O P I C R E V I E W |
| ovince |
Posted - 05/08/2006 : 7:18:17 PM
hi everybody,
when plotting a dataset I see 4 data that I would like to exclude
from analysis. I would like to do it using Origin C or LabTalk. How to do it?
thanks
oliver
|
| 9 L A T E S T R E P L I E S (Newest First) |
| Mike Buess |
Posted - 05/10/2006 : 08:38:15 AM
Hi Oliver,
This shows how to create a fit curve with the stat object...
stat.reset();
stat.data$=dataset name;
stat.fitxdata$=wks_Xname; // X dataset for fit curve (column must already exist)
stat.fitydata$=wks_Yname; // Y dataset for fit curve (column must already exist)
stat.makeX.fitnpts=fit_points; // number of points in fit curve
stat.makeX.fitX1=lower_limit; // lowest X-value on fit curve
stat.makeX.fitX2=upper_limit; // highest X-value on fit curve
stat.makeX(); // create X values for fit curve
stat.lr(); // perform fit and create Y values
%A=stat.fitydata$;
lay -i200 %A; // plot fit curve in active graph layer
You can also use the nlsf object which will obtain the parameters you want (and many more), supports masking and can draw the fit curve.
Mike Buess
Origin WebRing Member
Edited by - Mike Buess on 05/10/2006 10:00:15 AM |
| ovince |
Posted - 05/10/2006 : 05:51:58 AM
hi,
What I am trying to do is to analyse a large number of data sets.
1. For each data set I should calculate Coeff of Linear Correlation and Standard Deviation.
2. The data should be plotted together with the fitted curve in order to analise it by eye.
3. If the fit is not good to remove some data.
For this task I have used lr.r and lr.stat because it automaticaly plots the fitted curve.
With stat feature and fitpoly I cannot find how to do it. Could you help me?
oliver
|
| larry_lan |
Posted - 05/10/2006 : 05:34:23 AM
Hi Olive:
Sorry, it should be
stat.data$=datasetname;
to specify a dataset.
If you using OriginC, maybe you can use fitpoly() function to do the fitting directly. like:
void test()
{
Curve crv("Data1_A","Data1_B");
double coeff[2];
fitpoly(crv, 1, &coeff);
}
Edited by - larry_lan on 05/10/2006 05:41:06 AM |
| ovince |
Posted - 05/10/2006 : 05:12:39 AM
hi,
fals alarm...it works with LT_execute? If the data is plotted in an active graph window, lr adds the data plot of the fitted curve to the graph window. This is not the case with stat. What to do?
oliver |
| ovince |
Posted - 05/10/2006 : 04:44:24 AM
Hi,
Do you know how to include these commands
stat.dataset$=datasetname;
stat.lr();
stat.r=;
stat.sd=;
into Orugin C? I have tryed with LT_execute() but does not work
oliver
|
| larry_lan |
Posted - 05/09/2006 : 07:54:06 AM
Hi Oliver:
If you want to do linear fitting, you can use the stat.lr() command, lr dataset command will not exclude the masked data.
stat.dataset$=datasetname;
stat.lr();
stat.r=;
stat.sd=;
Then you can get the proper correlation coefficient and standard deviation.
Larry
OriginLab GZ Office |
| ovince |
Posted - 05/09/2006 : 02:29:06 AM
hi,
I have masked 4 data in Origin C by typing:
LT_set_str("%R", nameForLT); //nameForLT is the dataset name
LT_execute("%R<16>=1");
LT_execute("%R<466>=1");
LT_execute("%R<1089>=1");
LT_execute("%R<1561>=1");
I tought it works because the bad data were colored into red. But comparing analysis with masked and unmasked data (correlation coefficient and standard deviation) one can notice that the data were NOT really excluded from analysis. What whould be the next step?
oliver |
| ovince |
Posted - 05/09/2006 : 02:08:06 AM
thanks Mike...it works
oliver |
| Mike Buess |
Posted - 05/08/2006 : 8:59:32 PM
Hi oliver,
You can mask/unmask point (row) i in a dataset with the LabTalk command dataset<i>=n, where n=1 (mask) or 0 (unmask).
Mike Buess
Origin WebRing Member |
