Efficiency analysis of library environments: determining optimum working conditions with iot sensors and machine learning

This study aims to develop an Internet of Things (IoT)-based system for monitoring and analyzing environmental conditions in university libraries. In the system, environmental factors such as sound levels, light intensity, temperature, humidity, air quality, and crowd density are collected through sensors and microcontrollers, with crowd density measured using camera-based image processing techniques. User experiences are evaluated through feedback obtained via surveys. The collected data have been analyzed using algorithms including K-Nearest Neighbors (KNN), Logistic Regression, Decision Tree, Random Forest, Support Vector Machines (SVM), Extreme Gradient Boosting (XGBoost), and Naive Bayes. Modeling has been performed for each environmental factor separately, and all data have been integrated for a comprehensive analysis. According to the analysis results, KNN showed the highest performance with an F1 score of 96.14% for sound levels, Random Forest achieved 74.70% for light data, 90.14% for air quality analysis, KNN showed 98.13% for temperature data, Random Forest performed with 40.46% for crowd density analysis, and KNN reached the highest performance with an F1 score of 99.04% for overall evaluation. The developed user interface provides real-time environmental analyses for each user by utilizing the weight files of the trained models, evaluating parameters such as indoor air quality, sound levels, temperature, and light intensity, and presenting the main factors that degrade environmental quality to the user. By integrating user feedback with environmental data, the model determines the overall suitability of the environment and provides a holistic analysis of the efficiency of the work environment. The results show that the comprehensive monitoring of environmental factors and the integration of user feedback play a critical role in improving library work environments. Particularly, the KNN algorithm, with its F1 score of 99.04%, demonstrated that the integration of data integrity and user feedback significantly improves the model's accuracy. This thesis transforms environmental monitoring systems beyond classical control mechanisms into a flexible, scalable, and field-applicable decision support system integrated with user-centered feedback. The proposed approach offers a robust, scientifically-based, and applicable roadmap for sustainable digital transformation policies aimed at improving the environmental comfort not only of library settings but also of all shared spaces.