regression on principal component analysis

128 Views Asked by At

I have done a PCA to get my principal components and now do a principal component regression. In the PCR the 1., 2. and 4. component are significant and the 3. component is insignificant. Can anyone explain how this can be, because the 3. component are explaining more variance than the 4. component and shouldn't therefore the 3. component be significant and not the 4 ?

1

There are 1 best solutions below

1
On

Perhaps this is just a small misunderstanding. You do the destillation of the PC based on the 4 independent variables. Here the coordinate-system, in which the four variables lay, and after the rotation of the axes to the PC-position, the third Variable explains more than the 4'th variable - of that common variance.

But now you use that set of coordinates to explain a further variable. Nothing is there which prevents that the fourth variable does more then the third in this respect.