Numerous problems in geosciences, epidemiology, traffic planning and crime research nowadays involve large amounts of spatial point pattern data recording event occurrence. In many such applications, a main problem of interest is to characterize the probability of event occurrence and its relationship with a set of covariates, considering spatial dependence of observations. Spatial Poisson point process models are commonly used for the analysis of point patterns, in which the intensity function is assumed to be a function of a number of covariates. In this study, we seek to develop a spatial varying coefficient model for point patterns with large data via regularization. We propose a computationally efficient algorithm for parameter estimations and establish theoretical properties of the proposed estimators. We illustrate the performance of our model via both simulation studies and real data examples.