A Classification Tool for Indoor Air Quality Management with Explainability

Ensuring optimal indoor air quality is paramount for human well-being, demanding vigilant monitoring and classification by trained personnel. While machine learning models can streamline classification processes, it’s crucial that their outputs are easily understandable for staff. In our research, we leverage a publicly available dataset and employ k-means clustering, utilizing the elbow method to identify the optimal number of clusters (k). We enrich the dataset by integrating an ‘Indoor Air Quality Index” column derived from the clustering outcomes. Subsequently, we employ a Random Forest model to thoroughly analyze the dataset, resulting in highly accurate classification outcomes. Additionally, we introduce a Python-based web tool that not only presents various statistical parameters and visual representations of the data but also showcases the classification results from the machine learning model. Moreover, it provides explainability plots using the SHapley Additive exPlanations (SHAP) method. We firmly assert that this tool serves as a valuable resource for managers and health and safety professionals, enabling a deeper understanding of the pivotal factors influencing indoor air quality.