Ransomware attacks pose a significant and evolving threat to financial institutions, demanding robust predictive frameworks to enhance Cyber Situational Awareness (CSA). This study evaluates the performance of five prognostic models: IBk (k-Nearest Neighbor), Naïve Bayes, Hoeffding Tree, SMO (SVM-based), and Logistic Regression, in detecting ransomware threats across three categories: Data Breach, System Compromise, and Service Disruption. Using metrics of overall accuracy, precision, and recall, the models were compared for their effectiveness in knowledge recognition and threat forecasting. Results indicate that IBk achieved the highest overall accuracy (82.61%), with balanced precision and recall across all threat categories, making it the most reliable model for comprehensive ransomware detection. Naïve Bayes and Hoeffding Tree demonstrated strong recall, supporting early-warning systems, while SMO was effective for high-confidence detection of breaches and system compromises but failed in service disruption prediction. Logistic Regression showed the lowest accuracy and inconsistent performance. The findings highlight that integrating knowledge recognition with prognostic analysis significantly strengthens CSA by enabling accurate perception, comprehension, and projection of ransomware threats. The study recommends adopting IBk as the core predictive model, complemented by recall-focused classifiers and ensemble approaches, to achieve proactive, data-driven ransomware defense in financial sector environments.