Categorizing bicycling environment quality based on mobile sensor data and bicycle flow data

Abstract

The bicycle is a healthy and sustainable transport mode due to its emission-free characteristics. To increase bicycle use, it is fundamental to provide bicycle-friendly environments. To better monitor bicycle environments, this study proposed the concept of bicycling environment quality (BEQ), which was defined by perceived satisfaction and conflict level. Data collection was conducted at 19 road segments in five sites located in Beijing, China. Then, speed-related and accelerationrelated bicycling behavior indicators (BBIs) were extracted from data collected using sensors on mobile phones, while bicycling environment indicators (BEIs), such as bicycle flow, were extracted from recorded data. Taking the BBIs and BEIs as input attributes, a two-level BEQ classification assessment model based on a random forest (RF) algorithm was constructed. The proposed RF-based classification assessment model was able to produce approximately 77.35% overall correct classification. The results demonstrate the feasibility of using GPS data in evaluating BEQ. In addition, a novel dockless bicycle-sharing system (DBS)-based framework for bicycle traffic monitoring is discussed, which is of great significance in the sustainable development of bicycles. This study provides a theoretical method for objective BEQ assessment. It can further be used by planners and road administrators to monitor and improve BEQ and by individual cyclists for optimal route choice.