Enter a series of values to compute the mean absolute deviation MAD = sum|x - xbar| / n, measuring average dispersion for forecast error and robustness checks.
输入多个数值(逗号或空格分隔),求平均绝对偏差。
| Item | Formula / Note | Source / Standard |
|---|---|---|
| Mean absolute deviation | MAD = (1/n) · sum |x_i - xbar| | ISO 3534-1:2006, Statistics - Vocabulary and symbols |
| Relation to standard deviation | For a normal population, MAD ≈ sigma · sqrt(2/pi) ≈ 0.7979·sigma | Geary RC (1936), The distribution of the mean deviation, Biometrika 28 |
| Median absolute deviation | MedianAD = median( |x_i - median(x)| ) | Tukey JW, Exploratory Data Analysis (1977), Addison-Wesley |
| Robustness | MAD is less sensitive to outliers than the standard deviation (robust scale) | Tukey JW, Exploratory Data Analysis (1977) |
统计指标、置信区间与样本量的测算。
平均绝对偏差 MAD=mean(|xᵢ−x̄|):各数据与均值绝对偏差的平均值,衡量数据离散程度。
MAD 对离群值不如标准差敏感(不平方),量纲与原始数据一致,更易直观理解。单位同原数据。