Readme file
SERIES C
 Applied Statistics
Evaluation of trace evidence in the form of multivariate data
C. G. G. Aitken and D. Lucy
Appl. Statist., 53 (2004), 109 - 122
The accompanying files are:
evidence-evaluation.txt- R source for example glass-data.txt
- Elemental composition data for 5 replicates on 62 panes of
glass kindly supplied by JoAnne Buscaglia of the FBI MVRA-functions.txt
- R source for the functions
The file evidence-evaluation.r is an R script file which when run will
load the data in
glass-data.txt.
The data in glass-data.txt are elemental composition measurements on four elements for 62 panes of glass with 5 replicates for each pane.
File evidence-evaluation.r splits the data into a control group (3 measurements) and a recovered group (3 measurements) for a pane of glass (with 1 measurement in common between control and recovered). It then calculates a likelihood ratio for the hypothesis that the control and recovered are from the same pane to the hypothesis that the control and recovered are from different panes, taking into account the within and between pane multivariate measurements.
It does this for all panes, so the output is a series of LRs between control and recovered where there is 1 where it is known the measurements are from the same pane, and 61 where it is known the measurements are from different panes.
The terms control and recovered are arbitary terms used to describe a possible situation where one set of replicate measurements has been found on an object at a crime scene, and the recovered on an object in association with a suspect. They can be given any other meaning deemed appropriate.
The output is a matrix called output which has a column indicating which pane donated the control measurements, a column indicating which pane is the recovered pane and a column of likelihood ratios.
R is open source software and is availible from:
http://cran.r-project.org/
Dr. David Lucy
School of Mathematics
The University of Edinburgh
James Clerk Maxwell Building
King's Buildings
Mayfield Road
Edinburgh
EH9 3JZ
tel: 0131 650 5086
extension: 505086
e-mail: dlucy@maths.ed.ac.uk
web: http://www.maths.ed.ac.uk/~dlucy/
|
