At this point, I hope you are already familiar with simple linear regression. In many cases, however, you may face problem that your data is not linear but have tendency to make a curve. Non-linear regression is a functional relationship that does not produce a straight-line in the scattered plot. Some non-linear curves can be transformed into linear regression. In this simple tutorial, you will learn how to transform some non-linear regression into linear regression.
Here is step by step on when and how to use curvilinear or non-linear regression:
- Firstly, you plot your data into scattered plot (XY type graph)
- Examine if there is any non linear relationship on the scattered plot
- Guess the model that relate X and Y and transform the model into linear model
- Compute the parameters and statistical fitness of the model
- Transform back the parameter to non linear model.
Preferable reference for this tutorial is
Teknomo, Kardi. Non-Linear Transformation for Regression. http:\\people.revoledu.com\kardi\ tutorial\Regression\NonLinear\