Spatial-Temporal Analysis of Urban Mobility using Taxi Dataset

Abstract

Abstract Ride-sharing services linked to mobile devices are innovative, on-demand transport services that aim to reduce the ill effects of private vehicles, such as pollution, congestion, etc. Ride-sharing is one of the emerging technologies which contains large data, and users with the same origin-destination and travel time are matched and share the ride. With the release of the Uber Movement Dataset for certain large cities, the spatial-temporal analysis of urban mobility using a taxi dataset is now possible. The present study used aggregated Uber trip data from 2016 to 2019 for New Delhi. This paper explores the applications of python -based techniques such as big data analytics, machine learning, and probabilistic programming for predicting travel time by utilizing the Uber Movement Dataset of New Delhi by taking several origins and destinations. Time Series Forecasting has been carried out with the help of ARIMA, Holt-Winters, Facebook Prophet, and the global model, which shows the difference between actual and predicted travel time. Spatial analysis for different wards is conducted to understand the variation in many trips. The results of this study will be helpful in urban planning and a better understanding of human mobility in New Delhi.