The variability in bicycle demand depends strongly on weather. This paper describes a ‘weather’ model that makes demand forecasting possible. The model is based on flow time-series of many years, collected at 16 cycle paths in the Dutch cities of Gouda and Ede. The model is bi-level. The lower level describes how cyclists value the weather. The upper level is the relation between demand and this weather value. The observations show that most cyclists value the weather in a similar way, but recreational demand is much more sensitive to weather than utilitarian demand. Most fluctuations are described by the model, but a significant fraction is still not covered. From a correlation analysis of the residuals, we conclude that about 70% of the remaining variation is locally constrained, and can therefore not be described by a generic model. However, about 30% of this variation is not driven by local effects. The cause of this variation is not yet known. Besides uncovering trends in cycling, the model can also be employed to evaluate the effect of cycling policy interventions, and to correct flow measurements as input in traffic models.
Stored only when both papers are in the corpus — 47 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.