| Summary: | 碩士 === 國立臺灣海洋大學 === 商船學系所 === 94 === In the past, Taiwan was a seamen-export country, well-known in the maritime community, because of seafarers’ high quality and good performance. However, the quantity of seafarers was decreased yearly with an unknown reason. Thus, this thesis is trying to find out what are the major factors of this issue. Several factors are derived from the paper reviews. From the point of view of methodology, all of the conventional prediction methods could not deal with the correlations between factors. Consequently, the artificial neural network (ANN) is a suitable approach to solve this problem without those disadvantages.
The thesis used different data normalizations method to construct the ANN models. Then, it chose training and testing error of mean square root and contribution graph analysis as the criteria to evaluate different models. Additionally, it constructed the regression model to examine the relations between the factors and the quantity of seafarers as well. According to the simulation, the optimal model has been derived to make a prediction. In the light of research, the min-max normalization method is better than the others. Besides, the regression model cannot be constructed, hence the prediction is invalid. Notwithstanding, the ANN model is fitted to this subject.
|
|---|
