Operational performance metrics are necessary to detect anomalous floors, equipment, and work-order categories in large commercial buildings. To this end, this thesis presents a method to extract operational performance metrics from computerized maintenance management systems (CMMS) and text-based tenant surveys. The method was demonstrated by using work-order logs and text-based tenant survey data gathered from four large commercial buildings in Ottawa, Canada. The analysis of CMMS data highlights the potential of decision trees, Sankey diagrams and association node networks to effectively visualize anomalies in building complaint patterns. Investigation of the text-based tenant surveys using established text analytics algorithms reveals that classifiers are more accurate for sentiment analysis than lexicon-based methods while both association rule mining and topic modelling algorithms successfully uncover key operational insights. Finally, a software tool mock-up was developed that combines the most impactful elements from previous work for building owners to visualize complaint patterns and maintenance workflows.