There are many natural conditions in the world, such as floods and earthquakes that cause major accidents. The actual question is how to anticipate and predict such events, which enables us to prepare ourselves in advance. We are interested in the advance classification of incoming calls to the smart city 112 public safety system for Ljubljana, the capital city of Slovenia, and the whole country. We used open Internet of Things data for Ljubljana and the whole of Slovenia for the years 2013-2016. Using different classification methods, different models were obtained and compared with each other. We have provided a report on how accurate the forecasts are and what classification result differences occur when different models or data of the capital city or the whole of Slovenia are used. The results obtained are a sound basis for further work on improving the classification in advance of incoming calls to the smart city 112 public safety system.