Low Error-Rate Temperature Prediction with Machine Learning Using LoRaWAN Data

The sustainability of energy resources in smart cities is one of the key research topics today. In this study, 8702 data packets, each of which containing pressure, humidity and temperature data, collected by using an indoor Long Range Wide Area Network (LoRaWAN) were processed with the CatBoost machine learning algorithm to predict and validate the temperature. The prediction results were evaluated using a regression model built with the CatBoost algorithm and metric analyses. According to results, temperature values were successfully predicted with an error rate of 0.59%.