This pedagogical project is based on Kim Larsen blog post Sorry Arima, but I'm going Baysian based on S. L. Scoot and H. Varian paper: "Predicting the Present with Bayesian Structural Time Series". This project compares the usual frequentist approach with the bayesian approach for time series forecasting. The frequentist approach uses the well-known Arima model. The bayesian approach uses the hidden markow markov chain model called Bayesian structural time serie.