Exploring Incompleteness in Case-Based Reasoning: A Strategy for Overcoming Challenge

Abstract

Data quality is a critical aspect of machine learning as the performance of a model is directly impacted by the quality of the data used for training and testing. Poor-quality data can result in biased models, overfitting, or suboptimal performance. A range of tools are proposed to evaluate the data quality regarding the most commonly used quality indicators. Unfortunately, current solutions are too generic to effectively deal with the specifics of each machine learning approach. In this study, a first investigation on data quality regarding the completeness dimension in the case-based reasoning paradigm was performed. We introduce an algorithm to check the completeness of data according to the open-world assumption leading to improving the performance of the reasoning process of the case-based reasoning approach.