Artificial intelligence for supporting forecasting in maritime sector

The importance of the time series of data has always been of great relevance. A main use of them is the prediction of the future values of the quantities of interest. On this purpose, a lot of models have been created so far, as AR, MA, ARMA, ARMAX, ARIMA and so on. In the last years, the interest on Artificial Intelligence and Neural Network has grown a lot and a lot of studies were conducted to enable their use in different fields. This paper has the aim to show the possibility to use a system based on Artificial Intelligence to analyze the time series of index and future on the chartering of ships in order to predict the future values of them. The Neural Network is trained with the data of the last 3 years and the results obtained have be compared with those coming from ARIMA model and Carbon Copy model. The first aim of this paper is thus showing if the Neural Network performs better than the other 2 models and what day (first, third or fifth) is the best for the prevision made. The second purpose of this paper is establishing if the knowledge of the trend of the quantity value influences the results. The Neural Network has been trained both with a bullish trend and a bearish trend, then the results have been compared to prove if setting the right trend improve the quality of the prediction.