Stellato, BartolomeoCunningham, Holly2021-07-302026-09-302021-07-302026-09-302021-04-062021-07-30http://arks.princeton.edu/ark:/88435/dsp01st74ct567https://theses-dissertations.princeton.edu/handle/88435/dsp01st74ct567We propose a data-driven approach to on-demand transportation routing as a linear program with rider pickup times as parameters, using differentiable optimization to compute gradients of some cost with respect to these parameters. This solution allows for routing optimization in real-time and is a significant improvement in efficiency to the mixed-integer formulation.application/pdfenPrinceton University Senior Theses