Anomaly Detection in low cost environmental sensor data using Machine learning
Abstract
In 2021, a system was deployed at the makerere university weather station. This system consisted of four sensors that record temperature and humidity. These sensors are the SHT, HTC,BME and HDU. In environmental monitoring, a reliable sensor is is very crucial. Therefore, we developed an LSTM model based on the LSTM encoder-decoder architecture to detect anomalous points in the temperature and humidity data recorded by the four sensors to determine the most reliable for deployment in Uganda.