ABSTRACT
Chromatography is a versatile analytical technique, the results of which have influenced knowledge in many areas of the sciences. The performance of any chemometric analysis, however, requires organizing of experimental measurements into a data matrix. Data missing completely at random are usually substituted with the column or row means, while data that reporting limit are often replaced by zeros or half of the respective reporting limit. The chapter provides a comprehensive overview of the main concepts for handling missing values and discusses One of the most popular single imputation algorithms for handling missing completely at random (MCAR) measurements is the iterative algorithm. The easiest way to handle the left-censored measurements, which is still recommended sometimes, is to replace them with half of the reporting limit. Substitution of a large number of left-censored values in data, however, was reported to produce biased model solutions.
