Abstarct: Spatial point pattern data has various applications in the real world. For example, identification of earthquake points and Study of different species plants. Intensity function is the main Property in these data analysis. Several statisticians suggest different Approaches to modeling of this function such as, Binomial point processes, Poisson and Cox. According to nature of these data and cox processes flexibility, in this study we considered Subcategories of these processes as known log gaussian cox processes. log gaussian cox processes act very flexible because of simple improvise of auxiliary variables in its inside. Classic deduction in Leggings processes need to complicated calculations to earn approximate of the likelihood function. So because of this, researcher’s usual approaches for fit intensity function by Leggings processes is Bayesian. Since of posterior distribution form in these processes isn’t clear, the usual tools to fit modeling is Algorithms of the Markov Chain Mont Carlo (MCMC). Performance of these algorithms need to Programming skills and is serious in Convergence and computation time. A replacement approach is using of Bayesian approximate method as known Laplace approximation (INLA). This approximation in compare with MCMC algorithms is very quick and its result is the same in accuracy. In this thesis, we use the approximation method INLA for fitting the log gaussian cox processes models. To illustrate the application of the theoretical topics, we analyze the data of the earthquake point pattern of the northwest region of Iran with the log gaussian cox processes model.