Transformations

[astone788]

Nice one yendall. Gotta prepare for anything. We already know VCE further maths is a sick joke (Futher Maths is a sick joke!). So we gotta be extra prepared.

[Yendall]

Nice one yendall. Gotta prepare for anything. We already know VCE further maths is a sick joke (Futher Maths is a sick joke!). So we gotta be extra prepared.

You hit the nail on the head there! Further maths can be a real pain.

[Stick]

So we now know that a reciprocal transformation is more powerful than a logarithmic transformation. But how do they compare to a squared transformation? For example, if your scatter plot looked like the left side of an inverted parabola, what are the features that would tell a y^2 transformation is appropriate?

[Yendall]

So we now know that a reciprocal transformation is more powerful than a logarithmic transformation. But how do they compare to a squared transformation? For example, if your scatter plot looked like the left side of an inverted parabola, what are the features that would tell a y^2 transformation is appropriate?

It depends on how extreme it is. If it looked like the left side of an inverted parabola, i'd say a logarithmic transformation of x would be best fitting to linearise.
A squared transformation on Y would be far too extreme, same goes for a reciprocal transformation of x.

Or maybe i'm wrong?