A Machine Learning Framework for Predicting Student Placement Outcomes

Abstract

The accurate prediction of student placement outcomes is an important task for academic institutions to improve career support services. One difficulty is that most of the traditional statistical methods are not equipped to handle the interplay between the academics, complex variables and demographics that often combine to produce a niched placement success. In this work, we present a robust Machine Learning (ML) model for predicting outcomes for student placements, based on a large publicly available Kaggle dataset. The pipeline includes the successful prepossessing of all machine learning models with systematic data pre-processing, exploratory data analysis (EDA), encoding of categorical feature and data scaling for better quality of data for input to the models. Several machine learning algorithms, including Decision Trees (DT), Logistic Regression (LR), Voting Classifier (VC), and other classifiers are trained and their performances are compared. The model’s predictive performance is enhanced through hyperparameter optimization and cross validation. The proposed voting classifier outperforms the existing traditional ML models in terms of accuracy, precision, and computational efficiency. Our results show that machine learning models greatly improve predictability of a student placement and could be a valuable tool for data-driven career counselling and institutional planning. This research illustrates the value of artificial intelligence (AI) in the educational landscape and provides a stepping stone for further development such as real-time predictions and integrating a wider range of features.