Jordan's Question from Facebook (About Regression)

[Sudharaka]

Jordan from Facebook writes:

Help please,

 

[Prove It]

Jordan from Facebook writes:

Help please,

If we assume that a model of the form [math]\displaystyle \begin{align*} y = A + B\,e^{x} \end{align*}[/math] is appropriate, then we note that if we have [math]\displaystyle \begin{align*} X = e^{x} \end{align*}[/math], then we have a nice linear equation [math]\displaystyle \begin{align*} y = A + B\,X \end{align*}[/math].

So it would help to start by evaluating [math]\displaystyle e^x [/math] at each point x, giving a new set of data which we can call X. Then perform a linear least squares regression for data set y against data set X.