In test set data (Fig. 5), data of type 1, 2, 3, 6, and 10 are commonly ship engines with an age of 5 years or more. These types showed the lowest RMSE of test set prediction results (Table 3, Table 4). In Section 3.1, the data of the initial part is said to reflect fewer data than the actual (because the data in this study only include the military direct maintenance workshop). In other words, it can be said that the data of the initial part is less reliable than other sections of data. In general, the warranty repair period of the ROK naval ship engine does not exceed 5 years. Type 1, 2, 3, 6, and 10 do not include data for the initial 5-year period with relatively low reliability. That is, the HS model proved better performance than the comparative model by showing a low RMSE in the test set data of all engine types with relatively reliable data. 5.3. Reflecting the qualitative knowledge Prediction can be improved in the presence of the qualitative knowledge, construction era of the new engine type, for example. This act of translating qualitative into quantitative knowledge could be justified by analyzing their relationship with the existing failure functions of five engine types. Fig. 9 and Table 5 give two interpretable results. First, the failure function is shifted down due to technical development. For example, type 5 engine which replaced type 2 engine, takes a similar form with its predecessor, the main difference being its intercept. Second, the period of engine design has a big effect on its failure results. Euclidean distance between each function (Table 5), indicates that engine pairs (4 vs 5) and (2 vs 3) are relatively close. This is in line with our knowledge that type 4 and type 5 are new engines while type 2 and type 3 are old (Type 4 and 5 engines were constructed after 2010 while type 2 and 3 engines were constructed in the 1990s.) Based on the qualitative knowledge on the closeness of new engine type with the existing engine types, posterior of\(\overset{\overline{}}{\alpha_{e}}\ \) and\(\overset{\overline{}}{w_{e}}\ \) of previous engine types could be used as a hyperprior for new ship’s\(\overset{\overline{}}{\alpha_{s}}\ \) and\(\overset{\overline{}}{w_{s}}\). Compared to using the original prior\(\text{Normal}(\ \overset{\overline{}}{\alpha_{0}},\ \sigma_{\overset{\overline{}}{\alpha}})\)and\(\text{Normal}(\ \overset{\overline{}}{w_{0}},\ \sigma_{\overset{\overline{}}{w}})\ \)from equation 1, this would give more accurate results as more prior knowledge could be reflected for the prediction.