Software-defined networking (SDN) has emerged as a technique for overcoming the fundamental difficulties of traditional distributed networks by separating the control and data planes, resulting in a network that is extremely dynamic and easy to control. SDN is a revolutionary way of constructing the network architecture; nonetheless, its centralized settings make it prone to attack. The proposed research focuses on applying machine learning (ML) model to enforce security in an SDN scenario, by using a unit network as an example of SDN design, therefore, mitigating the distributed denial of service (DDoS) attacks in SDN and improving the robust behavior of proposed ML models under attacks. In the proposed research, the initial recognition of abnormal activities on SDN is performed by integrating hierarchical multi-class classification (HMCC) with binomial logistic regression (BLR). The detection of the abnormal attacks on the SDN topology through the proposed algorithm allows action against the source of the attack and makes the network architecture more flexible. The HMCC-BLR strategy is developed for detecting abnormal attacks to increase the accuracy of minority classes with imbalanced data. The obtained results have shown better detection accuracy with 99.48% and detection rate of 99.57% higher than the existing methods like hybrid model of support vector classifier with random forest (SVC-RF) and ensemble online machine learning-based intrusion prevention system (OML-based IPS).