Comparing forecasts for tourism dynamics in Medellín, Colombia

Tourism is a topic of interest to many economies around the world, but it has received limited attention in Colombia. Knowing the periods of larger tourist inflows is important for predicting coverage in services for tourists. In this paper, we compare the estimation between classical and Bayesian regression in order to choose the best alternative to predict the number of tourist arrivals to Medellin. We also identify the most significant variables affecting the influx of tourists and the models providing better fit to the associated dynamics. According to our results, the Bayesian approach shows better estimates than the classic one. In addition, the variable month is significant to explain the demands for both Colombians and foreigners. The periods with the highest incidence of visits to the city are December-January and June-July, a pattern that repeats itself every year, which is crucial for planning hotel resources.