Showing papers by "Jerome H. Friedman published in 1999"

Bump hunting in high-dimensional data

Abstract: Many data analytic questions can be formulated as (noisy) optimization problems. They explicitly or implicitly involve finding simultaneous combinations of values for a set of (’’input‘‘) variables that imply unusually large (or small) values of another designated (’’output‘‘) variable. Specifically, one seeks a set of subregions of the input variable space within which the value of the output variable is considerably larger (or smaller) than its average value over the entire input domain. In addition it is usually desired that these regions be describable in an interpretable form involving simple statements (’’rules‘‘) concerning the input values. This paper presents a procedure directed towards this goal based on the notion of ’’patient‘‘ rule induction. This patient strategy is contrasted with the greedy ones used by most rule induction methods, and semi-greedy ones used by some partitioning tree techniques such as CART. Applications involving scientific and commercial data bases are presented.

Cr-Pyrope Garnets in the Lithospheric Mantle. I. Compositional Systematics and Relations to Tectonic Setting

Abstract: Chrome-pyrope garnet is a minor but widespread phase in ultramafic association with Mg. The position and slope of the lherzolite trend vary with temperature and tectonic setting, suggesting that the P/ rocks of the continental lithosphere; its complex chemistry preserves a record of events related to fluid movements in the mantle, including T ratio exerts a control on Ca/Cr in lherzolite garnets. Garnets with less Ca than the lherzolite trend (‘subcalcic garnets’) are largely melt extraction and metasomatism. We have examined the majorelement and trace-element composition of >12 600 Cr-pyrope confined to Archon suites, where they typically are concentrated in the 130–180 km depth range. The few subcalcic garnets from (Cr2O3 > 1 wt %) xenocrysts in volcanic rocks to evaluate their compositional ranges and interelement relationships. Samples have Proton suites typically are lower in Cr and occur at shallower depths (100–120 km). Subcalcic garnets are absent in Tecton been divided into three major groups (Archon, [2·5 Ga; Proton, 2·5–1 Ga; Tecton, <1 Ga) depending on the age of the last major suites analysed in this work. The complexity of the geochemical relationships illustrated here, and their variation with temperature tectonothermal event in the crust penetrated by the host volcanic rock. Relative depths of garnets within each sample have been and tectonic setting, suggests that it is possible to define meaningful compositional populations of garnets, which can be used to map determined by measurement of Nickel Temperature ( TNi). Mn, Ni and Zn contents of Cr-pyrope garnets are controlled by T-dependent the stratigraphy and structure of the lithospheric mantle. partitioning between garnet and mantle olivine. The expected correlation of mg-number with T is largely masked by effects of bulk composition and crystal chemistry. The Cr content of garnet is a primary indicator of the degree of depletion of the host rock; Fe, Y,