| 释义 |
kurtosis Statistics.|kɜːˈtəʊsɪs|
[mod.L., f. Gr. κύρτωσις a bulging, convexity, f. κυρτός bulging, convex.]
A shape characteristic of a frequency distribution that reflects the sharpness of the peak (for a unimodal distribution) and the shortness of the tails, and is generally measured by the quantity µ4/µ22 or its excess over 3 (µ4 and µ2 being the fourth and the second moments about the mean of the distribution).
1905K. Pearson in Biometrika IV. 181, I have already called β2-3 = η the degree of kurtosis. 1931L. H. C. Tippett Methods of Statistics ii. 28 There are several curves having the same standard deviation but varying kurtosis. 1952W. L. Gore Statistical Methods for Chem. Exper. ii. 16 The kurtosis is useful in determining if a frequency distribution differs from the normal error curve. The kurtosis of a normal distribution is equal to 3; smaller values than 3 indicate a flatter distribution than the normal (a platykurtic distribution), while values above 3 indicate a more sharply peaked distribution than the normal (a leptokurtic distribution). 1968P. A. P. Moran Introd. Probability Theory vii. 317 Tables of..the co-efficient of kurtosis µ4µ2-2-3. 1972P. Laslett Household & Family in Past Time iv. 129 These variables were then summed and averaged: measures of skewness and kurtosis were computed. |