Hello!
Using SciDAVis 2.3.0 for Windows, binaries downloaded from this site.
I just tried to linear fit a test dataset, and I can't get any output of it except 'NaN' for all parameters with message
Iterations = 0
Status = iteration is not making progress towards solution
I'm aware of this and this, but I'm not trying to use weights at all (the setting is "weights unknown"). Moreover, nothing changes whatever option(s) I try to use, including passing a set of nonzero "instrumental weights".
What I'm doing wrong? Thank you.
Are your data defined as 'numeric' or as 'text' in the table you used to create the plot?
You can check this in the 'Type' tab on the right side of the table (see attached figure).
Regards,
Sure, all columns are set as numeric. After your post, tried to set the format to "decimal" mode, the same outcome. However, I noticed that you've used "Linear regression fit" and I don't have such menu entry. In "fit wizard" I choose "Built-in"->"Polynomial", polynomial order = 1, then try to fit and it fails.
What I've missed yesterday is the "fit using built-in function" checkbox. If checked, fit converges to correct result. What I was doing wrong is using "add expression" and then not altering the "parameters" entry. The parameters were a,b and the function was defined (after "add expression" as a0 + a1 * x, of course it would fail with mismatching names.
TBH interface could be either a bit more friendly or foolproof :)
Thank you very much again!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The 'Linear regression fit' is the same as 'Fit Linear', which is under the menu Analysis -> Quick Fit -> Fit Linear. This menu entry is visible when your graph window is in the 'first plane' (I don't know another way to describe it).
If the window in the 'first plan' is a table, you'll not be able to access all fit options, but just the Fit Wizard.
Last edit: Fellype 2020-11-17
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello!
Using SciDAVis 2.3.0 for Windows, binaries downloaded from this site.
I just tried to linear fit a test dataset, and I can't get any output of it except 'NaN' for all parameters with message
I'm aware of this and this, but I'm not trying to use weights at all (the setting is "weights unknown"). Moreover, nothing changes whatever option(s) I try to use, including passing a set of nonzero "instrumental weights".
What I'm doing wrong? Thank you.
Hi Suthiro,
Are your data defined as 'numeric' or as 'text' in the table you used to create the plot?
You can check this in the 'Type' tab on the right side of the table (see attached figure).
Regards,
Fellype
Sure, all columns are set as numeric. After your post, tried to set the format to "decimal" mode, the same outcome. However, I noticed that you've used "Linear regression fit" and I don't have such menu entry. In "fit wizard" I choose "Built-in"->"Polynomial", polynomial order = 1, then try to fit and it fails.
What I've missed yesterday is the "fit using built-in function" checkbox. If checked, fit converges to correct result. What I was doing wrong is using "add expression" and then not altering the "parameters" entry. The parameters were
a,b
and the function was defined (after "add expression" asa0 + a1 * x
, of course it would fail with mismatching names.TBH interface could be either a bit more friendly or foolproof :)
Thank you very much again!
Suthiro,
The 'Linear regression fit' is the same as 'Fit Linear', which is under the menu Analysis -> Quick Fit -> Fit Linear. This menu entry is visible when your graph window is in the 'first plane' (I don't know another way to describe it).
If the window in the 'first plan' is a table, you'll not be able to access all fit options, but just the Fit Wizard.
Last edit: Fellype 2020-11-17