The Hidden Cost of Real-Time: Quantifying Performance and Economic Trade-offs Between Streaming and Batch Machine Learning Across Domains

Nantale, Shamsa Halima Kasozi (2025) The Hidden Cost of Real-Time: Quantifying Performance and Economic Trade-offs Between Streaming and Batch Machine Learning Across Domains. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Streaming machine learning offers real-time insights, but how much performance do we sacrifice compared to traditional batch processing? This research quantifies this trade-off for the first time across two contrasting domains: credit card fraud detection with 0.167% fraud rate and electricity price prediction with 42.5% minority class. The study compared batch algorithms against streaming alternatives, testing three enhancement strategies to improve streaming performance. Results show streaming completely fails on fraud detection achieving 0% F1 score without enhancements. Even with improvements, streaming reaches only 79.2% of batch accuracy while costing 63.2% more due to excessive false alarms. On balanced electricity data, streaming works reasonably well with just 7.7% accuracy loss. The stark contrast reveals that data imbalance doesn’t hurt streaming gradually but exponentially, as a 440-fold difference in imbalance creates an 8-fold difference in performance loss. For businesses handling rare events like fraud, batch processing remains the better choice despite sacrificing real-time capability. This work helps practitioners choose between streaming and batch based on their data characteristics and cost constraints.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Horn, Christian UNSPECIFIED
Subjects:	H Social Sciences > HG Finance > Credit. Debt. Loans. Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	02 Jul 2026 14:02
Last Modified:	02 Jul 2026 14:02
URI:	https://norma.ncirl.ie/id/eprint/9440

Actions (login required)

View Item