ANN analysis of some social and economic factors on post-disaster population growth: The case of August 1999 Gölcük Earthquake

Disasters date back to earlier times than human history. People who survive in the region where disasters occur migrate to other regions due to factors such as security, shelter, employment, health concerns and psychological problems. It is an important national security problem for the disaster region, which is already negatively affected by the devastating effects of the disaster, to regain at least its pre-disaster social and economic position. In this study, the effects of some social and economic parameters (Socio-Economic Development Index (SEDI), the number of newly built independent sections for residential purposes, agricultural area, the number of small and large cattle, the number of health personnel and GDP variables) affecting the population of Gölcük after the 7.4 Mw earthquake centered in Gölcük on August 17, 1999 were analyzed by Artificial Neural Networks (ANN). The relevant coefficients were obtained for the selected nonlinear activation function including tangent hyperbolic and logarithmic functions. The reliability of the obtained results was increased by performing ANN analysis twice. Quite high performance values (Rmse, Mad, R2 etc.) were found from these analyses. Then, the effect of the above variables was examined by using Activation function. The variables that had the most positive effect on the Gölcük population were obtained as GDP with 269% and then the number of newly built independent sections for residence with 91.2428%. On the other hand, the variables that had the most negative effect were determined as SEGE with -97.7802% and livestock statistics with -80.4924%.