Kurtosis removal for data pre-processing

Abstract Mesokurtic projections are linear with null fourth cumulants. They might be useful data pre-processing tools when nonnormality, as measured by the cumulants, is either an opportunity or a challenge. Nonnull cumulants opportunities extreme kurtosis used to identify interesting nonnormal features, for example clusters and outliers. Unfortunately, this approach suffers from curse of dimensionality, which may addressed projecting onto subspace orthogonal mesokurtic projections. challenges using statistical methods whose sampling properties heavily depend on cumulant themselves. ease problem allowing use inferential same under normality. The paper shows necessary sufficient conditions existence compares them other gaussianization methods. Theoretical empirical results suggest that transformations particularly finite normal mixtures. practical illustrated AIS RANDU datasets.