Akademik Veri Yönetim Sistemi
9th International Artificial Intelligence and Data Processing Symposium, IDAP 2025, Malatya, Türkiye, 6 - 07 Eylül 2025, (Tam Metin Bildiri)
In Türkiye, which is an earthquake country, the most anticipated earthquake is the Istanbul earthquake. The biggest problem of an earthquake expected to be 7 and above is the damage it will cause. The purpose of this study is to make approximate estimates of the number of injured, dead, and demolished buildings in a possible Istanbul earthquake and to raise awareness. First, earthquakes that occurred around the world and in Türkiye were obtained and analyzed. Based on the analysis results, the regression method was first considered, but the models did not give satisfactory results due to the skewness of the data. Therefore, a classification was performed and the results were divided into nine separate categories, and ranges were obtained for the number of dead and injured people and the amount of the collapsed building. All models were tested by combining the earthquake data in Türkiye and the earthquake data worldwide. As a result, the Random Forest model gave the best results in estimating the number of dead and collapsed buildings for different magnitudes of earthquakes, and the LightGBM model gave the best results in estimating the number of injured people. The Pix2PixGAN model was used to create an earthquake simulation in Istanbul. The model, which was trained with the before and after images of earthquakes that occurred in Türkiye - mainly 6th February earthquakes has predicted the potential outcome of an earthquake in Istanbul. The results obtained for all models are presented to users within the framework of a website.