Feature Drift-Aware Boosted Ensemble (FDABE) for Concept Drift Adaptation in Industrial Sensor Dataset

Purpose: This study develops a Feature-based Drift-Aware Boosted Ensemble (FDABE) framework for effective concept drift adaptation in industrial sensor data streams. The objective is to identify and measure feature-level drift in dynamic industrial data streams and rigorously test the framework against baseline models to verify improvements in long-term classification accuracy and overall system robustness. Design/Methodology/Approach: This study uses the UCI Gas Sensor Array Drift Dataset, which contains 13,910 samples collected from 16 sensors over 36 months. Data were obtained from reputable sources such as Springer, ScienceDirect, Elsevier, and IEEE Xplore. Machine learning methods such as Adaptive XGBoost and Random Forest were used to evaluate the proposed model based on accuracy and macro F-1 score. Research Limitation: The study was validated using a single industrial sensor dataset characterised mainly by gradual drift. The framework requires further evaluation under abrupt, recurrent, and complex drift conditions and across diverse industrial datasets. Findings: The framework was only tested on one industrial dataset where the data changed very slowly. The evaluation was conducted on a single industrial sensor dataset characterised primarily by gradual drift; therefore, FDABE's effectiveness under abrupt and recurrent drift conditions requires further validation. Practical Implication: FDABE can support reliable industrial monitoring and predictive maintenance by identifying drifting sensor features and selectively adapting the model, thereby reducing the need for complete model retraining and maintaining predictive performance under changing operating conditions. Social Implication: The proposed framework maintains stable predictive performance under changing environments by improving predictive maintenance, decision-making, and industrial monitoring reliability in Industry 4.0. Originality/Value: FDABE introduces a fresh way to handle shifting data streams by tweaking only the changing variables instead of retaining the whole model. Ultimately, the industrial system's accuracy and reliability increase significantly.

Authors

Publication Details

Journal
African Journal Of Applied Research
Published
2026-11-18
DOI
https://doi.org/10.26437/s34ejd61
Primary Topic
Data Stream Mining Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Feature Drift-Aware Boosted Ensemble (FDABE) for Concept Drift Adaptation in Industrial Sensor Dataset

Abhay Anil Dande, M. A. Pund, Amruta Deshmukh
3 links
African Journal Of Applied Research
Data Stream Mining Techniques
article

Feature Drift-Aware Boosted Ensemble (FDABE) for Concept Drift Adaptation in Industrial Sensor Dataset

Abhay Anil Dande, M. A. Pund, Amruta Deshmukh
article en

Abstract

Purpose: This study develops a Feature-based Drift-Aware Boosted Ensemble (FDABE) framework for effective concept drift adaptation in industrial sensor data streams. The objective is to identify and measure feature-level drift in dynamic industrial data streams and rigorously test the framework against baseline models to verify improvements in long-term classification accuracy and overall system robustness. Design/Methodology/Approach: This study uses the UCI Gas Sensor Array Drift Dataset, which contains 13,910 samples collected from 16 sensors over 36 months. Data were obtained from reputable sources such as Springer, ScienceDirect, Elsevier, and IEEE Xplore. Machine learning methods such as Adaptive XGBoost and Random Forest were used to evaluate the proposed model based on accuracy and macro F-1 score. Research Limitation: The study was validated using a single industrial sensor dataset characterised mainly by gradual drift. The framework requires further evaluation under abrupt, recurrent, and complex drift conditions and across diverse industrial datasets. Findings: The framework was only tested on one industrial dataset where the data changed very slowly. The evaluation was conducted on a single industrial sensor dataset characterised primarily by gradual drift; therefore, FDABE's effectiveness under abrupt and recurrent drift conditions requires further validation. Practical Implication: FDABE can support reliable industrial monitoring and predictive maintenance by identifying drifting sensor features and selectively adapting the model, thereby reducing the need for complete model retraining and maintaining predictive performance under changing operating conditions. Social Implication: The proposed framework maintains stable predictive performance under changing environments by improving predictive maintenance, decision-making, and industrial monitoring reliability in Industry 4.0. Originality/Value: FDABE introduces a fresh way to handle shifting data streams by tweaking only the changing variables instead of retaining the whole model. Ultimately, the industrial system's accuracy and reliability increase significantly.

African Journal Of Applied ResearchVol. 12(8)
Industry, innovation and infrastructure
Data Stream Mining Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon