Weight analysis of influencing factors of fishing capacity of marine fishing vessels using machine learning algorithm

Previous quantitative analysis is often made at the macro level, such as the fishing capacity of marine fishing vessels. There are some limited requirements on the number of indicators in the fishing vessel operation. In this study, a weight evaluation model was presented on the influencing factors in the fishing capacity of a single vessel using machine learning. Fishing monitoring data were about 200,000 rows from 2018 to 2019 in three provinces of the South China Sea. First, the cleaning of original data was implemented using quartile, principal component analysis, data standardization, and unique thermal coding, where reliable data of more than 40,000 rows was obtained. Secondly, machine learning was used to construct the BP neural network, decision tree, and random forest models. At the same time, the grid search and cross validation combined with the traversal cycle were used to create 6,000 generations of learning curves. The results showed that the random forest model performed the best in terms of mean square error, mean absolute error, and determination coefficient, where the determination coefficient of the best parameters group was 0.951, indicating that the random forest model was obviously superior to others. Finally, the weights of each index were extracted using the random forest, thereby obtaining the weights of fishing monitoring data. The result showed that the weights of various influencing factors were as follows: Output of nets(50.070%), PCA (after reducing the dimension of power, gross ton and length)(23.779%), trawls (including single tow, double tow and shrimp tow nets)( 9.409%), number of nets(6.782%), operating time(4.578%), gill nets(2.019%), net drawing(1.347%), seine nets(1.228%), cover nets(0.628%), fishing gear(0.122%), fishing tackle(0.022%), age of vessel(0.009%), material of fishing vessel (steel)(0.002%), material of fishing vessel (FRP) (0.002%) and material of fishing vessel (wood) (0.002%).The research results clearly represent the impact proportion of various factors, which can provide important technical support and reference for the quantitative evaluation and supervision of the fishing capacity of marine fishing vessels, ship reduction and conversion, renewal and transformation and other marine fishing industry management.