Abstract In recent years, an increasing number of ships have been equipped with sensors and monitoring devices to track power consumption across various onboard hotel systems, resulting in a significant increase in the volume and availability of operational data. However, due to equipment faults, transmission errors, sensor malfunctions, and environmental interference, such as vibrations and harsh weather conditions, the operational data may contain erroneous data points that are critical to assess and address prior to conducting data analysis. If these issues are not addressed, they can undermine the accuracy of the analysis and limit the reliability of the insights derived from the data. In this paper, a systematic approach to data preparation for analyzing electrical power consumption of hotel operations onboard a passenger ship is presented. This approach addresses the unique challenges posed by a complex dataset comprising over 18 million datapoints, which includes power consumption information of five onboard hotel systems, weather data, and operational parameters. This study employed a comprehensive approach to missing data imputation, utilizing techniques such as K-nearest neighbor (KNN) imputer, backward/forward fill, and linear interpolation, with each method specifically tailored to the unique nature of missing data within the dataset. The data preparation strategy also includes outlier detection, data smoothing, feature engineering and normality test to guide appropriate correlation analysis, with the goal of identifying relationships between power consumption and other parameters within the dataset for effective feature selection. The final result is a dataset free from distortions and unwanted anomalies yet preserving the integrity and characteristics of the original data and ensuring high data quality without compromise.
Okonkwo et al. (Sun,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: