Curve fitting a) define intersect b) fit without pre-defined fittype

2 次查看（过去 30 天）
Laura Iten
Laura Iten 2017-5-12
评论： Laura Iten 2017-5-12
Hey guys
I have some data (let's say X and Y) and I think it fits probably best to a 'power1' fittype. I have used the curve fitting toolbox as well as fitted programmatically of which I prefer the latter (because I have many datasets and I don't particularly like clicking buttons).
Two questions:
a) I know my model has to go through (0,0) and through (1,1). How can I force it?
b) Even though I think it'll be something like y= c*A^z, I would like something like "free fitting" where just the bulk of data is taken into account and fitted. I would like to avoid giving a fittype. Is this possible?

请先登录，再回答此问题。

回答（1 个）

Jacob Mevorach
Jacob Mevorach 2017-5-12
If I was going to do this I would try a variety of fits for each I thought might be good and I'd calculate the r^2 value for each and then go with the fit which had the maximum r^2 value.
  1 个评论
Laura Iten
Laura Iten 2017-5-12
I did that. But I wanted to go beyond defining a fittype already. Also, I would've liked to know which points would be considered as outliers under the "free fit" conditions.
Do you happen to have an idea how to solve a)? Fit through 0,0 and 1,1 ?

请先登录，再进行评论。

类别

在 Help Center 和 File Exchange 中查找有关 Linear and Nonlinear Regression 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by