Measuring the performance of field-services through support vector machines (SVM)

Today, most practices used for capacity planning and operations management in field service systems are outdated and overly simplistic. This paper presents an innovative method for the development of a performance management system (PMS) to plan and control the work of large and dispersed field force. We rely on data mining techniques such as support vector machines (SVM) to predict the time required to perform different kinds of field interventions. We have applied SVM techniques on a large amount of real data that we purposively select, extract and elaborate from the database of a case study company, to develop a regressive model. We then use this model to predict the target (expected) performance of each field technician and customer job, and to compare them with the achieved (actual) ones, in order to detect anomalies, organisational flaws, opportunistic behaviours, and any out-of-control situations that may require detailed analysis from the field service managers. Although this approach has been validated using data from a company that provides field services in the multi-utility industry, we believe that the application of SVM algorithms can bring benefits in manifold contexts.