Predictive Modeling to Predict the Residency of Teachers Using Machine Learning for the Real-Time

Abstract

To support the identification system of demographic features of educators, residential place is an essential feature and the machine learning techniques play a vital role to predict the residential place of educators based on their responses provided. This paper shows the predictive model to predict the residency of Indian university teachers concerning Information and Communication Technology (ICT) awareness with three algorithms Support Vector Machine (SVM), Multilayer Perceptron (MLP), Random Forest (RF). There are four different experiments are conducted on the primary dataset with 344 instances and 35 features. These Machine learning (ML) algorithms used three different testing procedures like training ratio, k-fold Cross-Validation (CV) and Leave one out. A statistical t-test is also applied to compare the time of prediction by each algorithm. The consequences of the first three experiments shown that the RF algorithm outperformed others in the prediction of the residency of teachers. The findings of the fourth experiment revealed that SVM’s CPU user time significantly differs from others. The authors recommended the RF model to be positioned as a real-time model cause of the highest accuracy of 72.8% with moderate prediction time 0.12 s.