Clustering multivariate time-series data from indoor environmental sensors is challenging due to data complexity and noise. This paper compares feature-based, raw data-based, and model-based clustering methods to identify the most effective approach for extracting meaningful patterns. We evaluate these methods using Silhouette Score, Davies-Bouldin Index, and Calinski-Harabasz Index. Results show that model-based methods, especially Gaussian Mixture Models, outperform others in cluster compactness and separation, despite some negative Silhouette scores. Feature-based methods provide moderate stability, while raw data-based approaches yield mixed results. Limitations include zone-level data without inter-zone links, absence of benchmark datasets with labels, and a limited range of tested algorithms. Future work will address these by expanding data scope, incorporating benchmarks, and exploring more clustering techniques. This study offers a concise comparison of clustering strategies for indoor multivariate time-series data, aiding future research in environmental monitoring and smart buildings.