Building ESexpert: The Machine Learning System for Voice Interactive Interview Score Assessment for Student Loan Funds

Abstract

Scholarship applicants' interviews generally last from an hour to an entire day. Several critical situations of each student caused the interviewer's feelings, such as stress, sadness, and anxiety. Listening to and discussing negative points can cause mental health problems. In order to decrease the duration of interviewing, This research paper discusses the development of a machine learning model for evaluating Student Loan Fund interview scores and an automated voice-based interactive machine learning interview scoring system. The process consists of building a machine-learning model using the Random Forest Algorithm with an accuracy of 99.54% and the SMOTE method to adjust the data asymmetry. The software was developed as a web application based on the SDLC for a second process, running Cloud Speech-to-Text of Google Speech API. Experts in all five areas evaluated the software, and the evaluation score was high. The research also looks at the Software Development Life Cycle (SDLC) models and the use of cross-validation and bootstrapping for accuracy estimation and model selection.