Predicting Candidates For Fit And Proper Test Using K-Nearest Neighbor

摘要

Existing condition for Fit and Proper Test using filtering method based on criteria, result of this method are many filtered candidates hence needed times to doing the test. In other side, allocation time should suitable by maturity level of both candidates because examiner is a basic management till high management. The method to measure predicting candidate is not exist, then this research using K-Nearest Neighbor. K-Nearest Neighbor is the method able to predicting based on data that measuring the closest distance with successful sample or unsuccessful sample that have a position. To measure the distance, the method using criteria such as grade, assessment, competency certificates, talent score, education, last position term, alignment with new position. KNN predicting 5 employee who have True Positive (actual is success then prediction is Success), it can reduce candidate more 40 percent. From Common Measurement explained that K=1 more accurate and precision than K=3 and K=5 but compared with True Positive and False Positive K=5 have best than K=1 and K=3.