Given f = x'yz + xy' + yz'
g = x'yz + xy + x'yz'
Let us simplify g :
g = x'yz + xy + x'yz'
= x'y(z + z') + xy
= x'y + xy
= y(x + x')
= y
So the function only reduces to y..Hence it is not functionally complete..
Now f = x'yz + xy' + yz'
= y(x'z + z') + xy'
= y(x' + z') + xy'
= x'y + xy' + yz'
Keeping z = 1 and y = 1, we have :
f = x' (1) + x (0) + 1 (0)
= x'
Hence NOT gate can be realised by the above substitution..
Likewise , keeping z = 0 , we have :
f = x'y + xy' + y
= x'y + y + x
= y + x
Hence OR gate is realised using the above substitution..
Now we know that { OR , NOT } is functionally complete ..And as we are able to reduce the given function to this set of known functions which is a functionally complete set of functions , hence f is functionally complete as well..
Hence f is functionally complete but g is not functionally complete..
EDIT : But one thing to mention here is in hardware level , for implementation of OR and NOT as shown above , we have taken help of 0 and 1 to achieve the functionality of OR and NOT functions..
Hence f is not fully functionally complete..