This study’s purpose was to identify and predict anomalous behavior in cloud computing infrastructure through analyzing computer metric data and software application performance metrics. An unsupervised learning method was used to dynamically learn about the feature metric data captured from the computer infrastructure. Using these related metrics, anomalous behavior was identified, such as degraded performance, application failure, absence of computer network connectivity, or hardware faults. Prediction of abnormal events provides substantial benefit to the business enterprise for business continuity, avoiding loss of business productivity, and avoidance of potential loss of sales.