This paper presents a variation focused cluster analysis strategy to identify typical daily heating energy usage profiles of higher education buildings. Different from the existing cluster analysis studies which were primarily developed using Euclidean distance as the dissimilarity measure and tended to group the daily load profiles with similar magnitudes, Partitioning Around Medoids (PAM) clustering algorithm with Pearson Correlation Coefficient-based dissimilarity measure was used in this study to group the daily load profiles on the basis of the variation similarity. A comparison of the proposed strategy with a k-means cluster analysis with Euclidean distance as the dissimilarity measure was also performed. The performance of the proposed strategy was tested and evaluated using the three-year hourly heating energy usage data collected from 19 higher education buildings in Norway. The results demonstrated the effectiveness of the proposed strategy in identifying the typical daily energy usage profiles. The identified typical heating load profiles provided the information such as the peaks and troughs of the daily heating demand, daily high heating demand period and daily load variation. The identified profiles also helped to categorize multiple buildings into different groups in terms of the similar energy usage behaviors to support further energy efficiency initiatives.