MAD-Based Robust Scores¶
What it is¶
PQC includes a robust outlier detector based on the median and median absolute deviation (MAD), used to reduce sensitivity to a small fraction of extreme points.
Definitions¶
For residuals \(y_i\):
Robust z-score used in PQC:
Why PQC uses it¶
Unlike mean/std z-scores, median/MAD remains stable when a subset of points is contaminated, making it a useful complementary detector.
Interpretation¶
larger \(|z_i|\) indicates stronger deviation from robust center
points are flagged if \(|z_i| \ge z_{\mathrm{thresh}}\)
Assumptions and caveats¶
majority of observations are not outliers
MAD can be zero for near-constant or quantized data; then robust z-scores are undefined and detector may return no flags
Small worked example¶
If median = 0 and MAD = 2e-7, a point with residual 2e-6 has:
With threshold z_thresh = 5, it is flagged.
References¶
Hampel, F. R. (1974). “The influence curve and its role in robust estimation.” Journal of the American Statistical Association, 69(346), 383-393.
Rousseeuw, P. J., & Croux, C. (1993). “Alternatives to the median absolute deviation.” Journal of the American Statistical Association, 88(424), 1273-1283.