RAC Heat Map

[Updated 04-Sep-2019; previous version accessible at www.4n6chemometrics.com/database_archive]

The Speir Research Group in the Department of Forensic and Investigative Science at West Virginia University currently houses a database of over 2000 shoes. To date, 1300 shoes have been fully analyzed, identifying a total of 72,306 randomly acquired characteristics (RACs). These RACs have been classified based on outsole location and shape; shape classes include variable, linear and compact geometries. This webpage provides the user with access to an interactive heat map detailing RAC frequency in terms of location and shape for a normalized shoe.

In the following frequency table, numerical values in the shaded row remain constant regardless of the user's interaction with the heat map, displaying data associated with total RAC count for the entire database (regardless of cell location). Conversely, the three unshaded rows automatically update to display RAC count and frequency for individual cells (5 mm x 5 mm) selected by the user (please select a location/cell of interest using a left mouse click). As of September 2018, we also report chance association (co-occurrence and indistinguishability), modified by tread elements. For more information about data included in the table, please see the bottom of this page for further description.

Horizontal Cell # Vertical Cell #
- -


Variable Linear Compact Any Shape
All RACs in Database 28,236 21,363 22,707 72,306
All RACs in Cell - - - -
Chance of Finding RAC in Cell - - - -
Unique Shoes Contributing to RAC Count - - - -
Shoes with Tread Contact in Cell - - - -
Chance of RAC Co-Occurrence - - - -
Chance of RAC Indistinguishability - - - -
Visually Evaluated / Total RAC Pairs - - - -
Similarities & Likelihood Ratios

Few RACs   Map Key   Many RACs

Heatmap of 72306 RACs

All RACs in Database: Reports the total number of RACs in the database, irrespective of location.

All RACs in Cell: Reports the total number of RACs at a specific location.

Chance of Finding RAC in Cell: Reports the empirical frequency of finding a RAC at a specific location.

Chance of RAC Co-Occurrence: Reports the empirical frequency of selecting two shoes at random (with contact agreement of tread elements) and finding a positional match at a specific location.

Unique Shoes Contributing to RAC Count: Reports the number of unique shoes contributing to the RAC count. If the number of RACs in the cell is greater than the number of unique shoes contributing to this count, one or more shoes exhibited more than one RAC at a specific location.

Shoes with Tread Contact in Cell: Reports the number of shoes (out of 1300 possible) that had tread in contact with the ground at a specific location.

Chance of RAC Indistinguishability: Reports the estimated frequency of selecting two shoes at random (with contact agreement of tread elements) exhibiting co-localized RACs with sufficient similarity to be deemed visually indistinguishable (based on the upper 95% credible interval for the probability of indistinguishability).

Visually Evaluated / Total RAC Pairs: Reports the number of RAC pairs visually examined by analysts, and the total number of RAC pairs at a specific location.

Similarities & Likelihood Ratios: Redirects the user to secondary webpages that report the similarity of RACs with positional co-occurrence, originating from different outsoles.

Note on RAC Shape Classifications: Automated categorization of a RAC's geometry does not always agree with subjective judgments of shape classifications by human observers. Using the set of absolute classification rules, such disagreements may be encountered approximately 0.4% of the time when examining all online RAC images.

Note on Fourier Descriptors: All similarity scores were computed using RAC binary images. For illustrative purposes, each binary image is accompanied by a corresponding Fourier descriptor. Visual deviations in shape may be encountered approximately 0.2% of the time when comparing RAC tracings and their corresponding Fourier descriptors.

The content on this web site was funded through a grant from the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice (2013-DN-BX-K043). Neither the U.S. Department of Justice nor any of its components operate, control, are responsible for, or necessarily endorse, this web site (including, without limitation, its content, technical infrastructure, policies, and any services or tools provided).