# [Scilab-users] Analytic models for empirical 3D data

2 messages
Open this post in threaded view
|

## [Scilab-users] Analytic models for empirical 3D data

Dear all,

This is a tough one, and I'm aware it may not be directly related to Scilab, so sorry if it is off-topic. I post it anyway because it could lead in a future to a toolbox.

I'm looking for general methods for obtaining analytic models for empirical data such as a variable which depends on two independent variables (i.e., a double entry table). The aim isn't just to interpolate but to find a sort of analytical law that approximates the data. This has two steps:

1) Finding an analytic function, of two independent variables with as few parameters as possible that can likely approximate the data.

2) Optimizing the parameters for the best approximation to the data

2) Is relatively easy by least-squares or other optimization techniques. It is 1) the difficult part.

Any help as to literature or known mehods will be welcome

Federico Miyara

 Libre de virus. www.avast.com

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
Open this post in threaded view
|

## Re: Analytic models for empirical 3D data

Hi Federico

>It is 1) the difficult part.

Is it allowed to give answer outside of Scilab? I have successfully used Curve Expert for finding the most suitable analytical function.

... highly recommended.

Best regards,
Claus

On 24.09.2019 02:50, Federico Miyara wrote:

Dear all,

This is a tough one, and I'm aware it may not be directly related to Scilab, so sorry if it is off-topic. I post it anyway because it could lead in a future to a toolbox.

I'm looking for general methods for obtaining analytic models for empirical data such as a variable which depends on two independent variables (i.e., a double entry table). The aim isn't just to interpolate but to find a sort of analytical law that approximates the data. This has two steps:

1) Finding an analytic function, of two independent variables with as few parameters as possible that can likely approximate the data.

2) Optimizing the parameters for the best approximation to the data

2) Is relatively easy by least-squares or other optimization techniques. It is 1) the difficult part.

Any help as to literature or known mehods will be welcome

Federico Miyara

 Libre de virus. www.avast.com

```_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users
```

_______________________________________________
users mailing list
[hidden email]
http://lists.scilab.org/mailman/listinfo/users