Analysing bicycle route choice is highly relevant for transport planning and modelling, as it directly influences the design and optimization of urban transportation systems. Numerous studies have investigated bicycle route choice, providing insights into cyclists' preferences and behaviours. Although machine learning methods are increasingly applied in transport planning, there remains a notable gap in research focusing on their use in analysing and modelling bicycle route choice.
This study aims to assess the utility of machine learning approaches in comparison to established methods such as the multinomial logistic regression (MNL) model. Therefore, a comprehensive GPS dataset comprising 25,730 GPS tracks from 1,361 cyclists in Dresden, Germany, was used. The data was processed and enriched with network data to incorporate relevant factors influencing route choice (e.g., cycling infrastructure, slope, surface, safety, land use etc.). Subsequently, the processed data was used to (a) estimate an MNL model and (b) train various machine learning models, including support vector machines, decision trees, random forests, and neural networks. Furthermore, the effect strength and impact direction of independent variables, as well as overall model accuracy, were determined and compared across all models.
The results reveal that machine learning methods perform comparably well to the traditional MNL model. However, interpreting the results remains challenging. The normalized feature importance of ML models indicates similar relevance for different independent variables as the MNL coefficients. Additionally, the partial dependence of each variable generally suggests the same direction of impact (positive or negative). Combining both measures in a directed normalized feature importance suggests that ML models capture the influence of independent variables on route choice probability similarly to the MNL model. However, while ML models slightly outperform the MNL model in terms of accuracy (86%), their performance varies, ranging from 83% (random forest) to 89% (support vector machine).

A comparison of international research literature reveals that cycling behaviour varies across different contexts. It is influenced by age, gender, and trip purpose, as well as infrastructure supply (e.g., cycling lanes, surface quality), environmental factors (e.g., slopes, scenery), and operational conditions (e.g., traffic speed, waiting times at intersections). International studies show similar tendencies regarding these influencing factors. However, findings are often location-specific, as cycling behaviour is also influenced by local conditions such as city size and topography. Direct comparisons of international study results are challenging because data and methods often differ between study areas.
To investigate whether and to what extent cycling behaviour depends on city-specific or local factors, a comprehensive Germany-wide study was launched to assess the generalizability of these influences. The identified factors may also be relevant in other countries, making knowledge of city-dependent influences valuable for cycling planning in general.
This contribution presents initial results from the research project “Cycling Behaviour in Germany”. The study is based on a nationwide GPS dataset on cycling, comprising more than 8 million bicycle trips from approximately 3,000 German cities (2024). To analyse cycling behaviour, we classified cities based on characteristics such as size, topography, and bicycle share, incorporated additional secondary data (e.g., infrastructure supply), and examined how city characteristics impact cycling behaviour. Preliminary results indicate general patterns, such as a preference for smooth surface conditions and cycling infrastructure and an aversion to long distances and steep slopes. However, the strength of these influences varies by city. Differences and similarities also emerge in trip length, cycling frequency, speeds, and waiting times, which appear to be dependent on city-specific characteristics.