Effective management of cloud computing service levels (e.g. availability, performance, security) and financial risk is dependent on the cloud service provider's (CSP's) capability and trustworthiness. The invention applies industry cloud service level agreements (SLAs) with statistical and machine learning models for assessing CSPs and cloud services, and predicting performance and compliance against service levels. Cloud SLAs (ISO/IEC, EC, ENISA), cloud security requirements and compliance (CSA CCM, CAIQ), along with CSP performance (SLAs, cloud services) are analyzed via Graph Theory analysis and MCDA AHP to calculate CSP trustworthiness levels. CSP trustworthiness levels are input with CSP SLA content, cloud service performance measurements and configuration parameters into machine learning Regression analysis models to predict CSP cloud service performance and cloud SLA compliance, and enable model analysis and comparison. This can be used to determine which regression variables provide the highest predictive accuracy, enabling cloud service customers (CSCs) and CSPs opportunities for transparency, traceability and effective governance of cloud service levels, cloud services and management of risk.