Hello Everyone,
why regression takes only a linear form ? is it just bcoz it’s underlying Assumption of linearity ?? or why it can’t take an curvature or exponential form ?
You can introduce non-linearities by creating transformed variables for the model. The model needs to remain linear in the parameters (regression coefficients), though (ie, E(y)= B0 + B1X + B2(X^2) this would create curvature in the model by squaring the quantitative X and adding that transformation to the model).
The CFA Institute curriculum covers only linear regression.
Certainly, you can have other forms for the regression curve.
If you plot a number of points in a graph in Excel and tell it to add a trend line, you’ll see a variety of nonlinear regression possibilities.