Supplier disruption prediction

This dataset provides a comprehensive view of dynamic supplier disruption risk scoring, combining multi-source indicators and intelligent modeling outputs for multiple suppliers across a simulated 60-day time horizon. It has been designed to support research and experimentation in risk & context-aware supplier selection, digital twin-driven supply chain resilience, and hybrid decision-support systems that blend knowledge graphs (KGs), semantic embeddings, and simulation-based risk-mitigation policies. The dataset contains daily records from January 1 to March 1, 2025, for 10 suppliers. Each row includes temporal disruption signals and pre-computed disruption risk outputs based on a fused analytical model. This model integrates a knowledge graph (KG) based supplier assessment criteria score (kg_score), NLP based (natural language processing) semantic embedding score (Em_Score), and five external disruption indicators: Twitter alert presence (twitter_alert), weather disruptions (weather_risk), economic risk (economic_risk), geopolitical signal (geo_risk), and technology disruption (tech_risk). Additionally, the daily supplier downtime (downtime_hours) is logged as a measure of operational impact. The final column, high_disruption_p, represents the daily probability of high disruption, calculated using a logistic regression equation.