Traditional demand estimation tools were developed for product design instead of product service system (PSS) design. PSS is a new market structure where the focus is on selling the use of a product instead of the product itself. Demand estimation faces challenges when applied to PSS design including mis-estimation, not being quantifiably repeatable, or built from evidence .
This thesis examines two PSS Design methodology questions. First, what is the effectiveness of spatially-derived revealed preference data in estimating distributed PSS demand? Estimating binomial distribution parameters n (user population size) and p (user population product affinity) can predict demand in new situations for distributed product service systems. Plots of binomial parameters reveal a continuous surface over the PSS area that allow more accurate prediction of relative ridership levels at new PSS locations.
Secondly, this work examine how designers can compensate for situations where the PSS design environment has changed and limited user data is available to create demand estimations. This thesis hypothesis that publicly available socio-demographic and environmental variables can inform multivariable regressions that estimate the n and p Demand Surfaces outside of the boundaries previously constrained by available user data.
Together, the answers to these two questions provide an initial framework to estimate Revealed Preference demand for many types of PSSs. In the examination of both questions, the proposed approaches are tested by the 2015 Chicago Bike Share System expansion. The effectiveness of these approaches is shown through analysis techniques including Spearman’s rho, Pearson’s Coefficient, Monte-Carlo Sensitivity Analysis, and resource impact of algorithm implementation.