Internet of Things (IoT) is going strong and is used in almost all the fields. The advancement in IoT technologies provides scope to improve our day-to-day life, but with this advancement also comes the challenges and obstacles, and one of the most important challenges required to be addressed is security. Various IoT devices are susceptible to many kinds of IoT botnet attacks, one of which is the Mirai attack. There are numerous machine learning methods, both supervised and unsupervised, that are used to handle these issues. Here a hybrid approach is proposed to detecting botnet attacks. It combines both supervised and unsupervised machine learning methods, and helps to determine new incoming attacks in the traffic. To achieve this, we first build the unsupervised machine learning method, and its outputs are fed into the supervised machine learning method where they are classified into benign or attack data.