Smart cities are still at their beginnings. They add intelligence to the existing city systems, enabling more to be achieved with less. Artificial intelligence can be helpful in this domain. The result is applications with transparent real-time information that aid in making better choices. This can be useful for everyday cases, but also for many natural conditions like earthquakes, tsunamis, floods, and sleet, which cause major disasters. The question is how to set up the concept of a smart city so that the city would be more resilient to disasters and climate change. The problem that needs to be addressed is to forecast the daily number of incoming calls based on data from IoT sensor devices, which includes river flows and weather data. The proposal is made for a safe city system, which is a smart city subsystem. The data refer to Slovenia as a whole and consist of both balanced and imbalanced data. AdaBoostM1 classification algorithm is used. The given evaluation will help in a more detailed analysis and in preparation of a comprehensive concept for the blueprint of a smart city.