Can you use non-linear variables in a linear regression model?

1 Answer
Mar 6, 2016

We can use linear regression model in non-linear relations in cases, where they can be expressed in linear form.

Explanation:

Let us assume a function such as y=kx^a, where k and a are two constants.

By taking logs of both sides, we get logy=logk+alogx, which can be written as bary=bark+abarx, a linear relation between bary=logy and barx=logx.

As the relation is linear, we can now use linear regression model. Hence, we can do this in cases where the non-linear elation can be expressed in linear form.