This repository was published to share the complete implementation of discrete time Markov chain model whose main purpose is to calculate a fraud risk measure to clients who deal with digital channels. The mentioned calculation is based on Montecarlo simulation of the model.
Markov chains are a suitable way to model sequencial data either for continuous and discrete time. The model here implemented belongs to a discrete time-countable states space class; states who are involved (or not) with fraud state are countable; time could be continuos, but for the sake of save coding efforts was deemed discrete.