PREDICTING COMPANY ABSORPTION OF ON-THE-JOB-TRAINING STUDENTS USING LOGISTIC REGRESSION: TOWARDS ENHANCEMENT OF OJT PROGRAMS OF HIGHER EDUCATION INSTITUTIONS

Logistic Regression is a classification algorithm. It is used to predict a binary outcome given a set of independent variables. Higher Education Institutions sends qualified students for OJT at affiliating agencies before they graduate. The main purpose of this research is to predict whether the OJT student will be absorbed or not by the affiliating agency based on the OJT evaluation criteria which included gender, student personality, technical knowledge and skills, office work management skills, adviser rating, OJT final grade, GPA and company type. Logistic regression was used to create the model and make the predictions, generate the confusion matrix, and compute the classification report statistics such as precision, recall, f1- score, and the accuracy value including the resulting ROC curve. Results revealed that the computed precision is 0.80, recall is 0.80, f1-score of 0.80 and an accuracy value of 0.88, which means that the standard deviation of the errors is good enough and the predictions made is 88% accurate, thus, the variables used are significant predictors of OJT absorption agencies, HEIs can benchmark on these predictors for OJT absorption to affiliating agencies. The author recommended that the metrics used in the OJT evaluation of student trainees be retained, and additional measures on technical skills, student personality and office work management may be added in consultation with representatives of affiliating agencies to further improve the OJT evaluation criteria of HEIs, hence increasing the chance of company absorption. Further studies might be conducted to include criterions that were not on the existing attributes in consultation with representatives of affiliating agencies to further improve the OJT evaluation criteria and increase the absorption and employability of OJT students. Further studies may be conducted applying other algorithms to determine the strength of the predictors in OJT absorption to affiliating agencies