Result: Custom Machine Learning in Sensor-Driven Hotels

This report explores the integration of machine learning (ML) in the hotel industry, using Sundvolden Hotel as a case study to investigate operational potential of ML in sensor driven environments. The study introduces an experimental framework called Trained Machine Facilitation (TMF), which applies Industry 4.0 technologies to optimize operations and management, with an example within ventilation. Based on data from 19 Airthings sensors (1-year time span) and the hotel’s own conference schedule (3-year time span), three specific models were applied: Heimdal for outlier/anomaly detection, Munin for predicting sensor values based on expected attendee numbers from the conference schedule, and Hugin for forecasting attendee numbers from sensor data. The results demonstrate the transformative power of ML as a tool to enhance operational efficiency through real-time and historical data insights. Heimdal excelled in identifying anomalies/outliers and achieved an F1-score of 0.97 (97%), underscoring its reliability as a gatekeeper for sensor data. Munin, after thorough data cleaning and optimization, achieved a high predictive accuracy with an R-squared of 0.904 (90.4%). However, Hugin faced challenges in forecasting attendee numbers, underscoring the complexities of sensor-driven predictions and data accessibility. Among the algorithms tested, the Random Forest algorithm emerged as the most robust, excelling in both prediction and processing efficiency. Additionally, the report emphasizes the critical role of data harmonization, prompt engineering and preprocessing in order to achieve reliable and high-quality predictive outcomes in dynamic sensor rich environments. Finally, the report concludes by evaluating and discussing the broader implications of integrating machine learning into building management systems, emphasizing the importance of choosing appropriate models for specific data-circumstances. This study has been conducted in collaboration with the Norwegian energy-efficiency and facilitation company Energy ...