Data mining and knowledge discovery tools become one of the foremost research areas in the field of medical diagnoses. The aim is to classify large datasets into patterns that can be used to extract useful knowledge. For example, data mining techniques can utilize patient’s databases for automated medical diagnoses. The purpose is to achieve more accurate findings, speed up the diagnoses, and reduce the errors and mistakes occurred by human being. However, incomplete dataset or missing features values may affect data mining findings. The problem of missing features values is common in many applications, particularly, in medical databases. The process of treating unknown attributes values with the most appropriate values is a common concern in data mining and knowledge discovery. The process of constructing missing values is a vital process in most supervised and unsupervised data mining researches because it may affect the quality of learning and the performance of classification algorithms.