We described model for prediction of number of people that have different levels of drug addiction on the specified area. Prediction is based on factors of the macro and micro state of the society on the specified territory. Several macro factors can be estimated based on the official data sources, but level of interest to drugs is much easier to estimate based on the data from social media. Moreover official data sources cover people with strong physical drug addiction but social media cover people with light drug addiction forms. We showed estimations of factors values for strong addicted people but estimation of factors based on the data from social media as well as modeling with these factors is a future work.