The Internet of Things (IoT) is an important and a major key component to support human life. Sensors are the sensing devices that play an important role in IoT to detect environmental conditions. The significance of IoT is increasing due to the increase in the things connected to the internet using Internet Protocol version-6 (IPv6). Low-Power and Lossy Network (LLN) Routers used in IoT environment have limited processing, memory, and energy resources and hence it can't use the traditional routing protocols such as Open Shortest Path First (OSPF), Routing Information Protocol (RIP), Ad Hoc On-Demand Distance Vector (AODV), Dynamic Source Routing(DSR) etc. However, Routing Protocol for Low power and Lossy Networks (RPL) is one of the prominent protocols which are developed to overcome the above issues but it could be exposed to specific types of attacks such as decreased rank attack, black hole attack, sinkhole attack, selective forwarding attack and version attacks, etc. Hence, it is necessary to develop a mechanism to detect and prevent such attacks. In this paper, efficient machine learning based Intrusion Detection System for Internet of Things is proposed to monitor the network activities against attacks and to detect the intruders more efficiently. This proposed machine learning based detection model performs feature selection and classification based on two new algorithms proposed in this paper called genetic recursive feature selection algorithm and fuzzy k-nearest neighbor classifier and detects the aforementioned attacks with maximum accuracy and minimum false positive rate. This work is also investigated on other performance metrics like precision, recall and F1-score to show the performance comparison between this proposed work with genetic recursive feature selection and fuzzy k-nearest neighbor classification algorithms and the existing works on intrusion detection.