Here are some clustering methods with their advantages and how to implement them in MATLAB –
- Dynamic Time Warping (DTW) with k-Medoids: DTW measures similarities between time series with temporal shifts, making it suitable for aligning load usage patterns that are out of phase. Combining DTW with k-medoids clustering, which is more robust to noise and outliers, can effectively group similar load profiles. In MATLAB, you can implement this using the “dtw” function for distance measurement and the “kmedoids” function for clustering. Please refer to the following documentation for “dtw” - https://www.mathworks.com/help/signal/ref/dtw.html
- Hierarchical Clustering: This method builds a hierarchy of clusters using either agglomerative (bottom-up) or divisive (top-down) approaches, which is advantageous when the data has a nested or hierarchical structure. MATLAB's linkage function can be used to perform hierarchical clustering, and the cluster function can extract clusters from the hierarchical tree. Refer to the following MATLAB answer question from community to understand more about hierarchical clustering algorithm - https://www.mathworks.com/matlabcentral/answers/1726630-how-to-apply-clustering-time-series-data?utm_source=chatgpt.com
- Fuzzy C-Means (FCM): FCM allows data points to belong to multiple clusters with varying degrees of membership, which is useful when cluster boundaries are ambiguous. This method provides a more nuanced clustering outcome, especially when dealing with overlapping load patterns. MATLAB offers the “fcm” function to perform fuzzy c-means clustering. Please refer to the following documentation for “fcm” - https://www.mathworks.com/help/fuzzy/fcm.html
