polynomial regression

Página 1

if linear regression does not seem to fir a given data set, polynomial regression can be an option

h-theta(x) = theta-0 + theta1-x1 + theta-2 * x1^2 + ....

or it can be :

h-theta(x) = theta-0 + theta1-x1 + theta-2 * sqrt (x1)

in these cases feature scaling becomes very important as for x1 -> 1-1000 ; x1^2 => 1-1000,000