Wireless Sensor Networks (WSNs) can make mistakes because they are used in dangerous and unpredictable surroundings. When it comes to finding problems in WSNs, there are many obstacles to overcome. It is challenging to precisely identify and diagnose faults in WSNs due to the diversity of deployments and the limited resources of the sensors. To find a solution to this problem, supervised machine learning-based methods are used to analyze the data from sensors to detect common occurring faults such as offset, gain, data loss, and stuck at the sensor node level. All mentioned faults were induced in normal data taken from a trustworthy dataset. Furthermore, the simulation results show that the extra trees-based machine learning method worked better than the state-of-the-art machine learning models.