No abstract

China’s planned mega-cities contain hidden, informal housing markets. We analyse Shanghai’s ‘group rental’ market in which formal commercial and residential units have been illegally converted into extremely crowded dormitories. In 2016, we collected more than 33,000 online classified advertisements for beds in group rental apartments and find that this market serves a specific demographic with robust preference order patterns. Furthermore, groundtruthing fieldwork revealed that the scraped online data misrepresented the market. Therefore, we also collected a second set of ‘real’ market data for comparative analysis. The study highlights both the exciting possibilities and the limitations of using online content to study informality.

We present a systematic method for generating activity-driven, multi-day alternative activity patterns that form choice sets for stated preference surveys. An activity pattern consists of information about an individual’s activity agenda, travel modes between activity episodes, and the location and duration of each episode. The proposed method adjusts an individual’s observed activity pattern using a hill-climbing algorithm, an iterative algorithm that finds local optima, to search for the best response to hypothetical system changes. The multi-day approach allows for flexibility to reschedule activities on different days and thus presents a more complete view of demand for activity participation, as these demands are rarely confined to a single day in reality. As a proof-of-concept, we apply the method to a multi-day activity-travel survey in Singapore and consider the hypothetical implementation of an on-demand autonomous vehicles service. The demonstration shows promising results, with the algorithm exhibiting overall desirable behavior with reasonable responses. In addition to representing the individual’s direct response, the use of observed patterns also reveals the propagation of impacts, that is, indirect effects, across the multi-day activity pattern.