Can I use log-log/linear regression models by taking the natural logarithm of some variables?
Imagine I have a dependent variable $y$ and independent variables $x_1$, $x_2$, and $x_3$. Can I run the following regression:
$$\log(y) = \log(x_1)+\log(x_2) + x_3$$