Statistical sampling security methodology for self-scanning checkout

Abstract

A statistical basis for use in a self-scanning checkout system determines how many items to check in a shopper's shopping cart for incorrect or missing scans as well as which particular or types of items to check to determine if they were properly scanned, if the shopper is determined to be audited. The present invention does not audit every customer, but rather determines whether a given shopper or customer is to be audited on a given shopping trip based upon obtaining a minimum checkout loss for such customer. The methodology determines how many items to check for a given shopper as well as which particular items to check for that shopper. The following factors attempt to model the real world of shopping and may be considered, alone or in varying combinations, in determining the number of items to check for a particular shopping transaction: shopper frequency; queue length; prior audit history; store location; time of day, day of week, date of year; number of times items are returned to shelf during shopping; dwell time between scans; customer loyalty; store shopping activity and other factors. Using statistical decision theory for auditing policies a minimum loss per shopper transaction improves the security and reduces the labor of self-check out without being too intrusive to customers.