Dealing with Label Noise in Machine Learning Predictive Models in Financial Revenue Management: A Clustering-Based Approach

Authors: Pavan Mullapudi

DOI: https://doi.org/10.5281/zenodo.15207279

Short DOI: https://doi.org/g9fc5d

Country: USA

Full-text Research PDF File:   View   |   Download

Abstract: This research addresses the critical challenge of label noise in financial revenue management predictive models. Label noise—incorrect or inconsistent class assignments in training data—significantly impacts model performance in financial applications where prediction accuracy directly affects revenue. We present a comprehensive analysis of existing label noise handling methodologies and propose a novel clustering-based framework that challenges the common assumption of uniform noise distribution. Our approach leverages unsupervised clustering to identify and correct non-uniform noise patterns in financial datasets. When applied to cloud financial management using public data, our framework demonstrates an average precision improvement of 14.3% compared to traditional methods. The results confirm that addressing the non-uniformity of label noise is essential for building robust predictive models in financial contexts where data quality issues are prevalent but often overlooked.

Keywords: Supervised Learning, Label Noise, Machine Learning, Revenue Management, Cloud Computing, Financial Revenue Management

Paper Id: 232270

Published On: 2023-10-03

Published In: Volume 11, Issue 5, September-October 2023

Share this