Description
Carsharing offers an environmentally friendly alternative to private car ownership. However, carsharing providers face the challenging task of matching shifting vehicle supply with fluctuating customer demand to prevent related operational inefficiencies and ensure customer satisfaction. To date, researchers have improved existing relocation strategies and developed new concepts with the use of information technology tools. Still, current literature lacks research on optimization and implementation of user-based relocation solutions. The most urgent need currently lies in the development of algorithms to compute and implement effective incentives for user-based relocation. We address these needs by utilizing a design science research approach to develop an automated machine learning-based incentive computation solution for incentivizing user-based relocation. We use a survey of 274 participants resulting in 1370 individual data points to train an incentive computation model, which is then applied within a small-scale field test. Results suggest that the algorithm computes appropriate incentives.
Computing Incentives for User-Based Relocation in Carsharing
Carsharing offers an environmentally friendly alternative to private car ownership. However, carsharing providers face the challenging task of matching shifting vehicle supply with fluctuating customer demand to prevent related operational inefficiencies and ensure customer satisfaction. To date, researchers have improved existing relocation strategies and developed new concepts with the use of information technology tools. Still, current literature lacks research on optimization and implementation of user-based relocation solutions. The most urgent need currently lies in the development of algorithms to compute and implement effective incentives for user-based relocation. We address these needs by utilizing a design science research approach to develop an automated machine learning-based incentive computation solution for incentivizing user-based relocation. We use a survey of 274 participants resulting in 1370 individual data points to train an incentive computation model, which is then applied within a small-scale field test. Results suggest that the algorithm computes appropriate incentives.