Percentiles are the points in a set of numerical values corresponding to how far (as a percentage) they sit through the ordered list. For example, the 10 percentile is the point such that 10% of the data values are below and 90% above. There are special cases such as the median which is at the half way point, or 50 percentile, and the lower quartile and upper quartile that lie at the 25% and 75% poinrs respectively.
It can be useful to transform data into percentiles during data preparation as it effectively turns the data into a uniform distribution and reduces the undue influence of outliers. It can also be useful for long-tail distributions as it reduces the impact of the small number of extreme tail values.
Used in Chap. 7: page 94
Used in glossary entries: long-tail distribution, outliers, uniform distribution