[4H3-J-13] AI application: agriculture and forestry

[4H3-J-13-01] Daily fish catch forecasting for fixed shore net fishing using bayesian state space model

〇Yuya Kokaki¹, Yuka Horiuchi¹, Naohiro Tawara¹, Masayoshi Hukushima², Akira Idoue², Kazuo Hashimoto¹, Tetsunori Kobayashi¹, Tetsuji Ogawa¹ (1. Waseda University, 2. KDDI Research, Inc.)

Keywords:state space model, time series analysis, fixed shore net fishing, machine learning

The daily fish catch forecast is indispensable information that can support fishery workers with their decision-making for efficient operation. Machine learning generally performs well if big data are available. Those data, however, are not always available in developing fish catch forecasting systems. The present study proposes a fish catch forecasting method using a state space model that emulates the process of fixed shore net fishing. Experimental comparisons conducted using actual fish catches and meteorological data demonstrated that the proposed method yielded a significant improvement over general machine learning-based forecasting when only limited amount of data are available.