End-to-End Data Quality-Driven Framework for Machine Learning in Production Environment

TL;DR

The paper presents an end-to-end framework that tightly integrates data quality assessment with real-time ML operations via drift-aware MLOps, demonstrated in an ESR steel production setting. By combining adaptive drift detection, quantitative data quality scoring, and model lifecycle management, the approach achieves a $12\%$ performance gain ($R^2=94\%$) and a fourfold latency reduction, while maintaining model-agnostic applicability. A key insight is the strong, positive link between data quality and predictive accuracy, with moderate quality thresholds balancing data retention and performance. The work offers a practical, scalable path for deploying robust, time-sensitive ML in dynamic industrial environments and informs future extensions to broader data types and deployment contexts.

Abstract

This paper introduces a novel end-to-end framework that efficiently integrates data quality assessment with machine learning (ML) model operations in real-time production environments. While existing approaches treat data quality assessment and ML systems as isolated processes, our framework addresses the critical gap between theoretical methods and practical implementation by combining dynamic drift detection, adaptive data quality metrics, and MLOps into a cohesive, lightweight system. The key innovation lies in its operational efficiency, enabling real-time, quality-driven ML decision-making with minimal computational overhead. We validate the framework in a steel manufacturing company's Electroslag Remelting (ESR) vacuum pumping process, demonstrating a 12% improvement in model performance (R2 = 94%) and a fourfold reduction in prediction latency. By exploring the impact of data quality acceptability thresholds, we provide actionable insights into balancing data quality standards and predictive performance in industrial applications. This framework represents a significant advancement in MLOps, offering a robust solution for time-sensitive, data-driven decision-making in dynamic industrial environments.

Figures (14)

• Figure 1: Comparison of ML system architectures. (a) Traditional approach with offline data quality assessment. (b) Proposed integrated approach with continuous data quality monitoring and drift detection.
• Figure 2: Data quality scoring process across multiple data quality dimensions (DQ).
• Figure 3: Change point detection mechanism.
• Figure 4: Phases of the proposed framework.
• Figure 5: Initialization phase pipeline and associated system artifacts.