Sankhya: The Indian Journal of Statistics

1999, Volume 61, Series B, Pt. 1, pp. 202--227

NONPARAMETRIC REGRESSION FOR ANALYSIS OF COMPLEX SURVEYS AND GEOGRAPHIC VISUALIZATION

By

GERALD WHITTAKER Economic Research Service, Washington
and
DAVID W. SCOTT Rice University, Houston

SUMMARY. TThe Averaged Shifted Histogram (ASH) provides a very fast, robust method of estimating multivariate nonparametric regressions. The ASH algorithm is fast enough that large (500,000 observations) georeferenced economic and physical data sets can be analyzed interactively. Applications include linking economic models to physical environmental models, analyzing effects of alternative policies on water quality, and examining the effects of agricultural policy on agricultural land values. The results of ASH estimation are conveniently visualized with three-dimensional slices from $\Re^d$ using a geographic information system (GIS).

AMS (1991) subject classification.62D05, 62G07, 62H11.

Key words and phrases. Mapping; spatial smoothing; averaged shifted histogram; environmental economics.

Full paper (PDF)