| Author |
 Topic  |
|
|
WSABHV
Germany
2 Posts |
 Posted - 07/24/2003 : 09:32:46 AM
|
Hi,
I want to fit a dataset with 3 independ variables and 1 depend. The used database for the fit should be limited of row 20 .. 80 (for example). The comands
nlsf.databegin = 20;
nlsf.dataend = 80;
have only effect for the depend variable. The other 3 variables are shown with the whole range of the data. (dialogbox nonlinear fitting / dataset)
If the limitation begins at the beginning of the dataset (row 0 .. 30, for example) the results are correct.
I need a comand to set the databegin und dataend for the independ varibales.
Dietrich |
|
|
easwar
USA
1965 Posts |
Posted - 07/24/2003 : 10:29:37 AM
|
Hi Dietrich,
I could reproduce what you see - the begin and end ranges set by the nlsf.databegin and nlsf.dataend commands show only in the independent variable listing in the GUI, and not in the dependent variables. However, the fitting process works fine. The fitting is performed only over the data range specified by your script commands. This can be seen by looking at the fit curves/fit results.
Easwar
OriginLab.
|
 |
|
|
WSABHV
Germany
2 Posts |
Posted - 07/25/2003 : 03:09:17 AM
|
Hi easwar,
thank you for your quick answer. I think you have mixed up the dependent (normaly "Y") and the independent variables. Nevertheless I try to discribe the wanted method of the script with more details:
1. Design of the worksheet
The different parameters and stations are in the Columns
In the rows are the reading for every month.
2. Analyse
I want to fit different groups of parameters and stations
and
different periods.
In total there are more than 100 combinations.
One Solution is to use a second worksheet for only the fitting dataset. For every combination the needed dataset had to copy from the basis worksheet to the fitting worksheet. The fitting dataset than started always with row 1 => no problem
Faster and more elegant is the analyse without copy.
For the first fit starting in row 1 the result is ok. The second fit using row 11 .. 30 (for example) the result is wrong, because the dependent variable use row 11 .. 30 and the independent use 1 .. 20.
If I take the fitting dialogbox and connect the dataset manuell the result is ok.
Is there a dependence at what position the commands databegin and dataend are. I think not.
This is a part of the used script:
NLSF.dataStep = 1;
NLSF.setindep("u","data1_b");
NLSF.setindep("v","data1_c");
NLSF.setindep("w","data1_d");
NLSF.setdepend("y","data1_a");
NLSF.dataBegin = 63;
NLSF.dataEnd = 100;
I mean there must be a modification of the command setindep
e.g.: nlsf.setindep("u","data1_b",63,100)
or the command databegin and dataend
e.g.: nlsf.datbegin("u") = 63
In the labtalk manual and the online help of origin I dont found nothing in this way.
I hope you can help me.
Dietrich
|
|
|
| |
Topic |
|
|
|
NLSF.databegin