The attempt to take down a website or service by flooding it with requests from a large number of users is referred to as a Distributed Denial of Service (DDoS) attack. They pose a significant risk to any company providing services through the Internet. Communication between devices and the target system itself may be slowed or stopped during a DDoS. It causes problems with using the Internet, such as sending and receiving emails and using other online services. There are two main types of ML-based DDoS detection methods now in use: supervised and unsupervised. To identify DDoS using supervised ML, a substantial amount of labeled data must be readily available. On the other hand, unsupervised ML methods may identify threats by examining data from the network itself. These methods have difficulties because of the sheer volume of data that exists in network traffic, the lack of precision with which detections can be made, and the prevalence of false positives. In this work, we provide a semi-supervised online sequential ML method for DDoS detection. The accuracy rate is 97% when compared to other machine learning techniques.