Comparative Analysis of Classification algorithms for Classifying Psychotypes

Abstract

Machine learning is a subdomain of Artificial Intelligence that makes a machine learn with the help of data. Classification algorithms follow a supervised learning methodology which allows labels to be assigned to the observations so that unobserved data can be labelled based on the training data. This paper intends to study different classification (Supervised learning) algorithms with the help of the MBTI dataset. MBTI Test is a Meyers Briggs Type Indicator test which helps us to identify an individual based on one of the 16 personality types. 4 classification algorithms namely, k-nearest neighbours. Decision Tree, Support Vector Machine and Random Forest algorithm are implemented on the KPMI Dataset. The evaluation metries (accuracy, precision, recall and f1-score) related to each of the classification algorithms are measured. A comparison of the metrics is tabulated to throw light on the best algorithm for the given dataset. As per the MBTI test, an individual belongs to 1 of 16 personality types based on whether an individual is an extrovert(E)-introvert(I), sensing(S)-intuitive(N), thinking(T)-feeling(F) or judging(J)-perceiving(P). The MBTI dataset is visualized to know the psycho-type of the employees. The visualization helps to identify the highly and rarely found psycho type. It is visualized that in employees of the MBTI dataset, most of the psycho-types are satisfied with their jobs. In the future, this algorithm will help in identifying student personality type, related student behaviour analysis, and its predictions related to their career choice, and remedial measures for improvement in personality.