Anyone that has developed a system on past data, would split the data in half then create the proposed system then run that system on the other half of the original untouched data and compare results....