On assigning drivers for a home-delivery system on a performance basis

Consider a firm providing home-delivery of small parcels within a short period of time. The paper proposes a method to estimate the productivity of each driver on any fictitious route using linear regressions. The method separates the trip time in terms of the time spent: with customers, driving, and all other activities. Our tests show that productivity, measured in terms of number of customers visited per hour, varies across drivers. Even more, the same driver's productivity vary for peak and off-peak demand periods considerably. This methodology should enhance the reliability of the driver scheduling programs, since it reduces the error and bias of the productivity offered by the assigned set of drivers. It shall also improve the reliability of the delivery time promised to the customer.