ABSTRACT
This study proposes a method for classification based on a Pattern Trace Network and Principal Component Analysis. Using data from a public Senior High School, the proposed method was applied to forecast whether or not a Senior High School (SHS) student would qualify for admission into the university. The method’s performance and the performance of Logistic Regression were compared. The independent variables used in the study are gender, boarding status, program of study, and continuous assessment. 79% of the data was used as a training set to estimate the parameters of the models and the remainder was used as a test set to evaluate the performances of the methods. The stages of the modelling procedure are: grouping the data into smaller subsets, preprocessing, Principal Component Analysis, computation of mean vectors of new descriptors of the pre-defined classes in the data. Judging from the methods’ performances on the test set, both methods are equally good, with 78% overall predictive accuracy each.
In terms of specificity and sensitivity, the Proposed Method had 87% and 63% respectively while Logistic Regression had 84% and 68%. The study concludes that the proposed method is an efficient alternative to solving a classification problem whose implementation does not necessarily require a sufficiently large sample. The study also reveals that, as far as continuous assessment is concerned, marks obtained by a Senior High School Student in the third term of S.H.S. 1 is enough to predict whether or not he or she would qualify to gain admission into the university. The study recommends among other things that the proposed pattern trace network approach is an effective method for classification.
