Wednesday, October 5, 2016 at 12:00 noon, free!

John Karon has provided a summary of his lecture:

"Geochemists can use chemical composition to assess potential sources of some types of artifacts. First, natural raw materials are analyzed to characterize the chemical composition and variability of potential geologic sources. Then cultural artifacts are analyzed and their compositions are compared with the ranges of compositions from the sources. These types of classification problems are encountered by scientists in all disciplines, and statisticians have developed methods to solve these problems. Steve Shackley, a geochemist and obsidian specialist who has retired to Albuquerque, has a database of obsidian composition from more than 20 known sources in the Southwest. I used Steve’s data on five trace elements to evaluate three analytic approaches to distinguishing among these sources: principal components analysis, clustering methods (eight different methods), and recursive partitioning.

Archaeologists have used principal components and clustering for years; recursive partitioning (decision trees) is only beginning to be used. Recursive partitioning works quite well to distinguish among the obsidian sources, but the others fail badly. Jim Moore provided Steve’s analytic data and his source assignments for approximately 450 artifacts from seven sites in the Pojoaque Valley. Rather than using multivariate methods, Steve uses bivariate comparisons and a hierarchical decision process to arrive at his source assignments. I used the source recursive partitioning results to predict the sources of these artifacts and principal components analysis to evaluate the accuracy of Steve’s and my predictions, including investigating the disagreements between our predictions.

In my talk I will provide background on these methods, gliding lightly over the underlying theoretical basis for each, and I will present graphics and tables showing how these methods perform for this problem and the successful analysis strategy that resulted. No mathematical or statistical background required! These ideas are useful for other similar problems, including analyzing ceramics, chert, and osteological data."

The Brown Bag talks will take place at the Center for New Mexico Archaeology at 12:00 noon in the CNMA library. Seating is limited. Admission is free.

The Center for New Mexico Archaeology (7 Old Cochiti Road) is located off of Caja del Rio Road, across from Challenge New Mexico on the way to the Santa Fe Municipal Golf Course. Take 599 to South Meadows Road, continue through the traffic circle west along the Frontage Road to Caja del Rio Road. CNMA is on the left-hand side of the road and is the large building with white sail-like skylights on the roof.