Data Mining Techniques: An Approach for Recruitment Prediction Analysis of Undergraduate Engineering Students

Abstract

To accomplish data mining approach to predict and examine the most important feature essential for the recruitment of undergraduate students of Hyderabad and to compare the data mining algorithm to explore best suited algorithm. Data Mining ID3 Decision Tree computation is applied on dataset mainly on nine specific attribute which are most essential attribute to get recruitment and find out which attribute is most essential among all attributes for a student to get recruitment.The comparative machine learning modeling approach of K-Nearest Neighbor approach, Random Forest approach, Support Vector Machine approach and ID3 algorithmic is also performed to find the accurate matched algorithm for the study. Afterapplying data mining decision tree ID3 algorithm computation on nine specific attributes that is Academicgrade,Practical Knowledge,Skilled Certificates, Project accomplishment, Subject Knowledge in written exam and Interview, Fear in Written exam and interview, Communication skill in interview, Confidence in interview, Practical interest which are much relevant to get recruitment.This research found that “Skill certificate” attribute derive the highest information gain among all the attributes and it is a root node in resulting decision tree.After that the accuracy score result oftrained machine learning models that is K-Nearest Neighbor model, Random Forest model, Support Vector Machine model and ID3 model over test dataset is found. The accuracy score of K-Nearest Neighbor model is 50%,Random Forest model is 90%, Support Vector Machine model is 80% and ID3 model is 90%. In terms of accuracy score the Random Forest and ID3 model bothare resulted same accuracy score with 90% and performed successfully. Both are best suited algorithm for this research. This research is helpful to reduce the non-recruitment problem and beneficial for Engineering Institutions, students, corporate sectors. In future this work will further extended by more attributes and advance machine learning process.