kurtosis
UK[kɜːˈtəʊsɪs]US[kɜːrˈtoʊsɪs]
n
A measure of the "tailedness" of the probability distribution of a real-valued random variable, describing how much the shape of the distribution differs from the normal distribution.
Morpheme Breakdown
kurt
osis
kurt
curved, arched
osis
condition, state
Etymology
The term 'kurtosis' was coined in the late 19th century from the Greek root 'kyrtos,' meaning 'curved' or 'arched.' It entered the specialized vocabulary of statistics to describe the degree of curvature or peakedness in a probability distribution curve. The suffix '-osis' standardly forms nouns denoting a state or condition, thus framing kurtosis as the condition of curvature. While the original Greek root described physical curvature, its modern statistical application is an abstract metaphor, quantifying how distribution tails and peaks deviate from the bell-shaped 'normal' curve.
Analysis
Structure: kurt (curved, arched) + osis (condition, state)
kurt (from Greek κυρτός (kyrtos), meaning 'curved, convex, arched'); osis (a noun-forming suffix from Greek -ωσις (-ōsis), indicating a state, condition, or action).
Examples
A high kurtosis indicates a distribution with heavy tails and a sharp peak.
The analyst calculated the kurtosis to assess the risk of extreme outcomes in the model.
Compared to a normal distribution, positive kurtosis suggests a higher probability of outlier events.