Efficient deployment of Internet of Things (IoT) sensors primarily depends on allowing the adjustment of sensor power consumption according to the radio frequency (RF) propagation channel which is dictated by the type of the surrounding indoor environment. This paper develops a machine learning approach for indoor environment classification by exploiting support vector machine (SVM) based on RF signatures computed from real-time measurements. Results obtained demonstrate that the combination of received signal strength (RSS) and channel transfer function (CTF) yields a classification accuracy of 83.0% for identifying the type of the indoor environment.