From the course: 11 Useful Tips for Regression Analysis
Unlock the full course today
Join today to access over 22,500 courses taught by industry experts or purchase this course individually.
Fractional variables
From the course: 11 Useful Tips for Regression Analysis
Fractional variables
- [Instructor] Standard polynomial regression isn't perfect. Lower order polynomials offer a limited family of shapes whilst high order polynomials often fit the data poorly at extreme values causing edge effects or waves in your models. An alternative is using fractional polynomial which provide an extended family of curves using only a small set of predefined powers. Let me demonstrate this to you with an example. Let me draw for you a variety of functions using a standard quadratic polynomial with a predefined set of coefficients. This is what we get. It's not bad. There's a reasonable amount of shapes here. Now, let me repeat this with a fractional polynomial model of the same degree with this same predefined coefficient as before. This is what we get. There's a much wider variety of shapes here. And that is the advantage of fractional polynomials. So what fractional polynomials? Fractional polynomials differ from…
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.