The thesis discusses a mathematical model of subsidence on the surface in the excavation area of Premogovnik Velenje. Adequacy of the adjusted model was confirmed by comparison with the four and five parameter logistic function. Statistical analysis of data, which were obtained with epoch measurements of points, that are a part of four observational networks, was carried out with the Wolfram Mathematica software. Analysis encompasses fitting of each model to data based on nonlinear regression, statistical tests of estimated models and their parameters and the prediction of subsidence development. All models are a good fit. Most applicable model is the modified sigmoid function due to better analysis results, less sensitive parameters and reduced model complexity. A simple program code enables subsidence prediction, graphical display of subsidence dynamic and the time estimation of the next and final epoch measurement.