Machine learning to support citizen science in urban environmental management

Area of Science:

• Environmental science
• Computer science
• Environmental management

Background:

• Machine learning (ML) and citizen science (CS) are vital for environmental management.
• Integrating ML and CS presents opportunities and challenges, particularly regarding data validity and quality assurance.
• Urban litter is a significant environmental challenge requiring effective monitoring and management strategies.

Purpose of the Study:

• To demonstrate how ML can support CS by providing quality assurance for urban litter assessment.
• To evaluate the efficacy of ML models in predicting regulatory metrics typically assessed by experts.
• To explore the synergistic potential of integrating ML and CS in environmental management.

Main Methods:

• Quantitative data collected via CS were used to train and test five ML models.
• The ML models predicted a qualitative, site-specific, multiclass "Litter Index" score.
• Model performance was evaluated using accuracy, precision, recall, and F-1 scores.

Main Results:

• The XGBoost ML model achieved the highest performance, with scores of 0.98 for accuracy, precision, recall, and F-1.
• ML models demonstrated a reliable capability to complement CS assessments for urban litter.
• The study confirmed ML's potential to enhance quality assurance within regulatory frameworks.

Conclusions:

• The integration of ML and CS offers significant synergies for environmental management, particularly in urban litter monitoring.
• ML provides a robust tool for quality assurance, enhancing the reliability of CS data in regulatory contexts.
• This integrated approach has broad applicability to other environmental management domains beyond urban litter.