Code severity evaluation is a crucial part of software development, but traditional approaches frequently use narrow criteria to determine how common flaws are. This work presents a new metric and a cutting-edge model intended to offer a more accurate and insightful evaluation of severity levels. The suggested action greatly improves decision-making procedures where a precise severity assessment is necessary. The results show this innovative measure may be used to reliably evaluate severity by using a decision tree classifier to predict severity codes based on it. Additionally, the suggested model (NMSC_KNOW) predicts severity codes with greater accuracy, demonstrating its value as a reliable tool for software development.