Objective:

To develop an automated eCoaching system that provides personalized activity recommendations based on real-time weather data and individual preferences.

Approach:

• Data Collection: Collected weather data for 18 months from thirteen cities in southern Norway.
• Algorithm Development: Developed an algorithm to annotate, process, and classify weather data for tailored exercise suggestions.
• Performance Evaluation: Evaluated the classification performance using metrics such as accuracy, precision, recall, F1 score, and Matthews correlation coefficient (MCC).
• Interpretability: Used local model-independent interpretable explanations (LIME) to explain individual predictions.
• Ontology Framework: Represented information semantically using an Ontology framework to ensure reliable recommendations.

Key Findings:

• The decision tree classifier achieved an accuracy of 99.1%.
• The system provides accurate and contextually relevant guidance for physical activities based on weather conditions.
• Developed rules to determine appropriate activity types corresponding to different weather conditions.

Interpretation:

The eCoaching system integrates weather data and personal preferences to enhance physical activity recommendations.

Limitations:

• Current eCoaching systems often provide generic guidance and rely mainly on user activity history.
• Existing studies on weather and physical activity have limited use of contextual factors.

Conclusion:

The study addresses gaps in personalized eCoaching by integrating long-term weather data with personal preferences to generate context-aware recommendations.