Using Random Forest for Indoor Solar Energy Estimates on Embedded Systems

As IoT systems have gained a lot of interest in the last years, powering these devices have become a major hurdle. Energy harvesting is a solution to this problem but comes with its own difficulties. As the harvested energy is very scarce and big energy storages are costly and have an environmental impact, the energy usage have to be scheduled carefully with power availability in mind. It was suggested, that random forest is suitable for this task. In this thesis, I implement an energy prediction algorithm based on random forest on a microcontroller and evaluate its performance. Further a technique is implemented to improve models on the device itself.