Composite transformation using matrices, where did my approach go wrong?

Here is the problem: A square plate has vertices at A(1,1), B(-1,1),C(-1,3), and D(1,3). It is translated by 5 units along the +x-direction and then rotated by 45 degrees about P(0,2). Determine the final coordinates of the plate vertices.

My answer:

I will take point A(1,1) and transform it only.

Order of transformations:

First translate

then translate+rotate+translate-back

Initially,

This is the translation matrix

This is the A's coordinates matrix.

Now we multiply them and get

For rotation, we translate first by -2 in y-axis. we still get the same result of matrix.

then rotate by 45 degrees

After translating back to by +2 in y, we still get the same as above.

The rotation matrix that I have used is

Where did my approach go wrong?