The Intelligent Mobile Data Collection (IMDC) framework is presented in this study for use in IoT-based sensor networks. Clusters of nearby Internet of Things (IoT) devices and sensors are organised using this approach. A gateway node acts as the coordinator of each cluster and is responsible for gathering information from its members, processing it, and then sending it on to the main data collector (MDC). In order to examine trends in data creation, the framework uses a labelling method called FRL (Very Rare Labels). Time intervals, packet numbers, and types are some of the parameters that can be used to classify clusters using these labels. Local models that take into account states, actions, and rewards are developed by individual Internet of Things (IoT) sensors or devices within the FRL framework using reinforcement learning (RL) approaches. After collecting data from various regional models, the gateway compiles it into a global model and returns it to the IoT-Internet of Things sensors. In addition, the framework uses cluster categories to find the MDC's sleep length, visitation time, and Time Division Multiple Access (TDMA) slots, among other factors. This framework's efficacy in lowering latency and energy usage during data collecting and enhancing accuracy is demonstrated by its implementation using NS2.