I'm having a quite simple question: Why is a spline fit not the best choice everytime? In other words: How do I separate a spline fit from a kernel smoother or a polynomial in a meaningful way? I'm aware of the differences and I can imagine proper applications but imagine a case where you can use each of them. Like, for example, a quadratic curve with "some" noise.
$\begingroup$
$\endgroup$
5
-
3$\begingroup$ Generic answer is that every model makes a number of simplified assumptions about the reality and those assumptions fit some problems better then others. $\endgroup$– TimCommented Nov 2, 2018 at 9:28
-
$\begingroup$ Thank you! How I would apply this to the example of a quadratic curve? $\endgroup$– BenCommented Nov 2, 2018 at 9:40
-
3$\begingroup$ You are probably right that splines should be used much more but always is a strong word ... $\endgroup$– kjetil b halvorsen ♦Commented Nov 2, 2018 at 9:40
-
$\begingroup$ For some data sets a spline might not interpolate or extrapolate as smoothly as a single simple function, and there are some situations where this might be important. $\endgroup$– James PhillipsCommented Nov 2, 2018 at 15:05
-
$\begingroup$ thank you! But let's say we want to interpolate a guadratic behaviour: Why shall I use spline over kernel or a polynomial? The latter would probably the best but let us add some noise. How to select now? $\endgroup$– BenCommented Nov 2, 2018 at 15:25
Add a comment
|