Enhancing Conversions and Lead Scoring in Online Professional Education DOI: https://doi.org/10.33093/ijomfa.2024.5.1.2

Abstract

This study seeks to enhance lead conversion for online professional education providers by using supervised machine learning algorithms for lead conversion targeting and lead scoring, including Logistic Regression, K-Nearest Neighbors, Support Vector Machines, Naïve Bayes, Random Forst, Bagging, Boosting, and Stacking. A lead dataset was used to train and test the machine-learning models. The Recursive Feature Elimination (RFE) is used to establish a precise lead profile. The performance of the trained lead conversion models was evaluated and compared using the 10-Folds cross-validation method based on accuracy, precision, recall, and F1-score. The results show that Stacking is the best model with an accuracy of 0.9233, precision of 0.9391, and F1-score of 0.8939. Meanwhile, the Logistic Regression-based lead scoring model demonstrated promising potential for automating lead scoring. The results of the Logistic Regression-based lead scoring model achieved an accuracy of 0.9019, recall of 0.9019, precision of 0.9015, and F1-score of 0.9014. The optimal lead scoring threshold is 0.20, which stroked the optimal trade-off balance between accuracy, sensitivity, and specificity.