The worldwide outbreak of COVID-19 in February 2020 has had significant repercussions on the transport sector and mobility. Under the new enforced measures of restricted movement, it is crucial to understand the changes in mobility patterns. Based upon an online survey collected from more than 100 countries during the first wave between March – May 2020, the travel mode choice characteristics before and under the pandemic were investigated for three main trip purposes: commuting to work, educational trips, and shopping trips. This study explores the changes in mobility patterns on the global scale as well as for the case of Austria. Discrete choice models and machine Learning algorithms were applied to predict the change in travel behavior. The results imply a modal shift from the use of public transit towards privately owned vehicles, biking and walking during the pandemic first lockdown period. The accuracy of the two approaches fluctuated due to the uneven distribution of target labels in the dataset. Consequently, other assessment metrics such as the confusion matrix and the Matthews correlation coefficient were used to find out which model was the most suitable in predicting the modal choice. The results also show that machine learning classification algorithms have better performance than models based on random utility theory.     «

The worldwide outbreak of COVID-19 in February 2020 has had significant repercussions on the transport sector and mobility. Under the new enforced measures of restricted movement, it is crucial to understand the changes in mobility patterns. Based upon an online survey collected from more than 100 countries during the first wave between March – May 2020, the travel mode choice characteristics before and under the pandemic were investigated for three main trip purposes: commuting to work, educati...     »