The rapid expansion of digital services across e-governance, financial systems, healthcare platforms, and cloud-based infrastructures has significantly increased exposure to cyber threats. Reports published by CERT (Computer Emergency Response Time)-In indicate a steady rise in cyber incidents, including phishing, ransomware, distributed denial-of-service attacks, and unauthorized access attempts. Conventional signature-based detection mechanisms are increasingly inadequate against evolving and previously unseen attack patterns. This paper presents a data-driven framework that transforms system and network logs into predictive insights using machine intelligence. The framework is validated using secondary datasets aligned with national cybersecurity advisories and publicly available intrusion detection benchmarks widely adopted in academic research. Experimental evaluation demonstrates that ensemble learning models can effectively predict malicious activity with high accuracy, supporting proactive and automated cyber defense strategies.