Daily Ship Traffic Volume Statistics and Prediction Based on Automatic Identification System Data

Abstract

Daily ship traffic volume statistics and prediction are of great significance to shipping market. Reliable prediction of daily ship traffic volume can instruct shipping company to make sound judgment and decision for operational management. Because of the mobility of ships, it may be difficult to obtain the ship traffic volume automatically and frequently. This paper defines four types' daily ship traffic volume for a port and proposes autonomous statistic methods for counting daily ship traffic volume at port only based on Automatic Identification System (AIS) data. Take Shanghai port as instance, we count the daily ship traffic volume by using the proposed statistic methods for three common types of ship: cargo ship, passenger ship, and tanker ship. For ship traffic volume prediction, we apply a hybrid methodology that combines both Auto-Regressive and Moving Average (ARMA) and Artificial Neural Network (ANN) models to cargo ship and tanker ship, and show the comparison of ARMA, ANN and the hybrid methodology. Empirical results indicate that the hybrid methodology is efficient and superior for ship traffic volume prediction. But for passenger ship traffic volume, since the correlation coefficient plot shows feeble correlation in it, we just analyze some statistical characters for it.