Bio
Jing (Mirah) Zhang is a PhD student at the School of Geographical Sciences, University of Bristol. Jing holds Masters degrees in Quantitative Methods, Urban Analytics, and a BArch degree in Architecture. She also has extensive professional experience in the real estate sector in China and East Asia. Jing’s background enables her to think critically and creatively. Her research is uniquely interdisciplinary, as she draws inspiration from a broad range of intellectual resources from engineering to the social sciences and humanities. Jing currently works on causal theory and causal inference, with a focus on computational spatial causal models. This is an emerging field of methodological research at the intersection of statistical causal inference and machine learning. Jing’s research sets out to advance causal inference methods with the aim of advocating reliable analysis for scientific inquiries and decision support. Besides methodology development, Jing’s recent work during postgraduate studies engaged with policy sensitive topics such as housing, mobility, economic resilience and inequalities.
Research interests
Jing’s project at the Alan Turing Institute falls within several research areas, including statistical methods, causality, urban analytics. Her methodological work involves elements of causal discovery, counterfactual prediction, uncertainty quantification, information theory, and network theory. While a big part of causal inference is ‘normal’ statistics, Jing employs machine learning techniques where traditional statistical models are insufficient in handling causal relations with underlying spatial structures. Jing’s PhD project ‘Model based spatial causal inference’ is organised around two related problems in spatial causal settings: spatial interference and spatial confounding, both originated from the structured interaction of observational units. Jing’s project at the Alan Turing Institute consists of two methodology development tasks for her PhD: (1) ‘Causal measures across spatial scales’, where the goal is to explore methods for capturing causal relations systematically and automatically across spatial scales. (2) ‘Spatial causal interference with uncertainty over network structures’, which investigates how uncertainty in the interference network, as model input, affects downstream causal inference tasks. The application area for this research project is wide, as spatial structures exist in geographic proximity relations, infrastructure networks, economic input-output relations, as well as social networks.