Abstract
A method is proposed for the construction of a classification rule based on both complete and incomplete data records. This method incorporates the Hocking-Smith estimation procedure to find estimates of the mean vectors and variance-covariance matrices from samples which may contain incomplete data vectors. The resulting estimates are used to construct a discriminant function. We will demonstrate that these incomplete observation vectors contain information relevant to classification and should not be discarded. The proposed method is applied to the placement of students into one of two college algebra courses at Tarleton State University. The standard procedure that ignores incomplete observation vectors in the construction of a classification rule is also applied to the same data, and comparisons are made between the two methods. The proposed method that utilizes the incomplete observation vectors correctly classifies a higher percentage of students than does the standard method.
Get full access to this article
View all access options for this article.