Impact of imputation methods for ship technical parameters on emission estimations in ports

Greenhouse gas emissions from ships have emerged as a pressing concern. Nevertheless, the quality of data in existing databases remains inadequate, with numerous instances of missing information. This presents significant challenges for accurately estimating emissions associated with ship activities in port. This paper uses three imputation methods and applies them to three ports as a case study to evaluate their performance in emission estimation. The mixed-method demonstrates high accuracy while covering nearly all cases of missing data, resulting in the smallest error in estimating daily emissions. The results indicate that if the data quality is not improved, at least 12% of CO2 emissions may be underestimated. The cases of missing data that the imputation model can address also have a significant impact. For example, the multiple linear regression method, which only covers partial cases of missing data, leads to an underestimation of emissions by 2% to 6%. The findings highlight that an appropriate imputation method can significantly improve the accuracy of emission estimation. They also highlight the importance of data quality, which not only reduces estimation errors but also helps prevent the substantial underestimation of emissions.