Paper ID

2219

Description

In order to encourage individuals to adopt more sustainable behaviors, it is crucial to know their current levels of consumption in specific domains (e.g., mobility) before exposing them to personalized incentives. Although various theoretical models exist, there is currently no technological solution that automatically estimates individuals’ consumption sustainability levels. This short paper aims at addressing this gap and presents the design of a framework that enables to estimate these levels based on multiple features (e.g., demographics). It also presents a preliminary validation of a part of the framework through two empirical comparative studies related to the mobility consumption domain. These studies evaluate the performance of six classifiers using a large-scale survey of approximately 3000 representative individuals living in Switzerland. The results highlight that the gradient boosting trees and the multinomial logistic regression models are promising, and accommodation, habits and demographic variables are the most decisive features to estimate mobility behaviors.

Share

COinS
 

Supporting Green IS through a Framework Predicting Consumption Sustainability Levels of Individuals

In order to encourage individuals to adopt more sustainable behaviors, it is crucial to know their current levels of consumption in specific domains (e.g., mobility) before exposing them to personalized incentives. Although various theoretical models exist, there is currently no technological solution that automatically estimates individuals’ consumption sustainability levels. This short paper aims at addressing this gap and presents the design of a framework that enables to estimate these levels based on multiple features (e.g., demographics). It also presents a preliminary validation of a part of the framework through two empirical comparative studies related to the mobility consumption domain. These studies evaluate the performance of six classifiers using a large-scale survey of approximately 3000 representative individuals living in Switzerland. The results highlight that the gradient boosting trees and the multinomial logistic regression models are promising, and accommodation, habits and demographic variables are the most decisive features to estimate mobility behaviors.