Code covered by the BSD License  

Highlights from
Non-linear regression GUI

Be the first to rate this file! 25 Downloads (last 30 days) File Size: 32.8 KB File ID: #26159
image thumbnail

Non-linear regression GUI

by

 

This GUI solves model-based non-linear regression problems of the univariate and multivariate form.

| Watch this File

File Information
Description

The non-linear regression problem (univariate or multivariate) is easily posed using a graphical user interface (GUI) that solves the problem using one of the following solvers:
 - nlinfit: only univariate problems.
 - lsqnonlin: can deal with multivariate problems (more than one dependent fitting variable, ydata is a matrix).
 - patternsearch: this solver is useful to obtain a good start point, before using nlinfit or lsqonolin; this way, the global minimum is determined easier.

Data is introduced in the GUI as vector or matrix from the workspace.
The model to be fitted must be written in an M-file in vectorized form:
     ypred = model(x,xdata)
 ypred is a column vector (univariate problem) or matrix (multivariate problem) with the model response (observations in rows).
 x is a vector with the parameters of the model to be fitted.
 xdata is a matrix with the independent variables in columns and observations in rows.

An example of a kinetic equation fitting using data from a batch chemical reactor is enclosed.

Requirements: Optimzation toolbox/ Statistic toolbox/ Genetic algorithm and direct search toolbox depending on the selected solver.

Required Products Global Optimization Toolbox
Optimization Toolbox
Statistics Toolbox
MATLAB release MATLAB 7.8 (R2009a)
Tags for This File   Please login to tag files.
Please login to add a comment or rating.

Contact us