After reading the SEER LIMITED-USE RECORD DESCRIPTION, we found that some important attributes like ‘Tumor Size’, ‘EOD Extension’, and ‘Lymph Nonde Involv’ in the record of 1983-1987 should be obtained from the ‘4-Digit Extent of Disease’. So we divided it into three values and fill them into the corresponded filed. After this step, the records with missing information in the above attributes are removed from the data set.