Predicting Pedestrian Counts using Machine Learning

Abstract

Abstract. The study of urban population dynamics has long been an important area of research. In particular, the ability to accurately predict the number of pedestrians in a place and time is critical for urban management, population health, crime, and for quantifying the impacts of public events. However, it can be extremely difficult to analyse the size and characteristics of the ambient population due to limited data availability and difficulties in capturing non-linear relationships between pedestrian counts and external factors. This paper reports on an ongoing project that is using machine learning techniques to: (i) better understand the impact that the built environment and other contextual factors, such as weather conditions, will have on the size of the pedestrian population during the day and; (ii) predict the number of pedestrians under different conditions. The case study area is the city of Melbourne, Australia, where abundant pedestrian count data exist. Early results demonstrate that, broadly, the model appears to perform sufficiently well to be useful, and that modelling errors are not consistent across space or time (some times/places are easier to predict than others).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 6 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.