In recent years, with the changing pattern of holidays and the formation of short-term holidays, cities have found the opportunity for tourism development. Four types of the most important types of domestic tourism in Tehran, based on the statistics of the National Center of Statistics and the views of the experts in this area, is Medical, VFR, Recreational and Business tourism. For this purpose, the present study seeks to propose models for forecasting effective variables on forecasting domestic tourism demand in Tehran based on these four types. To do this, information was used between the years 2001 to 2015. Independent variable of this study is the number of domestic tourists in Tehran, based on these four types and dependent variables were selected based on Delphi and Fuzzy DEMATEL techniques. The model framework is a combination of regression, fuzzy neural network, and SVR algorithm, which combines these methods to measure forecast errors and compare the methods. The results of this research show that the proposed approach of regression can have better prediction than other methods for forecasting domestic Medical tourism and the proposed hybrid approach of regression and Adaptive Neuro-Fuzzy Inference System (ANFIS) can have better prediction than other methods for forecasting domestic VFR and Recreational tourism and the proposed hybrid approach of regression and SVR algorithm can have better prediction than other methods for forecasting domestic Business tourism.