Using XGBoost and memetic programming to identify hotspots of sediment plastic pollution

Despite growing global initiatives on sustainable plastic management, less than 10 % of plastic waste is effectively recycled, resulting in widespread environmental dispersion and pollution. This study examines the relative influence of topographic, hydrologic, and urban factors on the proliferation...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Environmental pollution (1987) Jg. 387; S. 127329
Hauptverfasser: Shiwomeh, Desmond N., Kantoush, Sameh A., Saber, Mohamed, Sumi, Tetsuya, Fantong, Wilson Y., Quang Nguyen, Binh, Abdrabo, Karim I., Mabrouk, Emad
Format: Journal Article
Sprache:Englisch
Veröffentlicht: England Elsevier Ltd 15.12.2025
Schlagworte:
Algorithms
Anthropogenic plastic hotspots
Boosting Machine Learning Algorithms
Cameroon
Environmental Monitoring - methods
Environmental Pollution - statistics & numerical data
Geologic Sediments - chemistry
Influence parameters
Memetic programming algorithm
Natural plastic hotspots
Plastic debris dispersal
Plastics - analysis
Proliferation of plastic hotspots
XGBoost model
Cameroon
Memetic programming algorithm
Anthropogenic plastic hotspots
Natural plastic hotspots
Plastic debris dispersal
Influence parameters
XGBoost model
Proliferation of plastic hotspots
ISSN:0269-7491, 1873-6424, 1873-6424
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Despite growing global initiatives on sustainable plastic management, less than 10 % of plastic waste is effectively recycled, resulting in widespread environmental dispersion and pollution. This study examines the relative influence of topographic, hydrologic, and urban factors on the proliferation of plastic hotspots (macroplastics) in the urbanized Mfoundi subbasin of Yaoundé, Cameroon. To achieve this, we employed Extreme Gradient Boosting (XGBoost) and Memetic Programming (MP) algorithms to classify both anthropogenic and naturally occurring plastic hotspots based on twelve spatially explicit parameters. This was then followed by assessing model performance through five key metrics: accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). Results reveal that topographic and hydrologic factors exert a stronger influence on hotspot formation than urban variables. Among the urban features, population density, road proximity, and waste management infrastructure were more strongly associated with anthropogenic hotspots, while land use exhibited limited influence overall. When multiple parameters were combined, model performance metrics were observed to improve significantly (≥75 % accuracy). The MP algorithm demonstrated more robust generalization across test datasets, whereas XGBoost exhibited signs of overfitting. These findings underscore the value of spatially explicit machine learning models for guiding targeted interventions in plastic pollution mitigation. [Display omitted] •Topographic features were the most influential drivers of natural plastic hotspots.•Combining multiple parameters enhanced classification accuracy above 75 %.•Memetic programming yielded more stable generalization than XGBoost.•Urban parameters had a stronger influence on anthropogenic than natural hotspots.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0269-7491
1873-6424
1873-6424
DOI:10.1016/j.envpol.2025.127329