A new modeling strategy for the predictive model of chub mackerel (Scomber japonicus) central fishing grounds in the Northwest Pacific Ocean based on machine learning and operational characteristics of the light fishing vessels

• DOI: 10.3389/fmars.2024.1451104
• Publication Date: 2024-10-14T04:51:05Z
• Authors: Haibin Han; Chen Shang; Bohui Jiang; Yuhan Wang; Yang Li; Delong Xiang; Heng Zhang; Yongchuang Shi; Keji Jiang
• Keywords: Scomber

The chub mackerel ( Scomber japonicus ) is one of the most influential small pelagic fish in Northwest Pacific Ocean, and accurate modeling approaches model selection are critical points predicting fishing grounds. This study investigated changes catches days on no moonlight bright (2014-2022) compared differences predictive performance between LightGBM RF models three datasets under two [those based light vessels operational characteristics (Approach one) those not Two)]. results were as follows: 1) Stronger intensity (e.g., full moon) can limit efficiency vessels, with years showing a trend higher percentage than percentage, i.e., resulted lower days; 2) Compared to Modeling Approach Two, one, achieved better dataset B, while both A B; 3) Overall, One more satisfactory prediction performance, optimal complete C improved from 65.02% (F1-score model, Two) 66.52% Two); 4) Under approach One) (LightGBM model), importance variables (no days) B (bright mainly centered environmental variables, CV, SLA, SSS being important A, DO, SLA B. provides scientific reasonable undertaking for research purse seine which conducive guiding fishermen select operating area time fishery accurately comprehensively realizing balanced development fisheries terms ecology economy.