Date of Award
Price Elasticity, Time series models, Prediction, Yellow cab demand, ARIMA, STARMA, VAR, STAR, LASSO-STAR, Spatio-temporal model, Regression, Weighting matrix
The spatio-temporal variations in demand for transportation, particularly taxis, are impacted by various factors such as commuting, weather, road work and closures, disruption in transit services, etc. Identifying the factors that influence taxi demand and understanding its dynamic provide planners with the information necessary to improve the transportation systems and also help drivers to reduce their vacant time.
This dissertation focuses on important factors affecting the demand. In the beginning, the impact of price changes on the demand is studied. Chapter One discusses how the seasonal effects and trends are removed from the demand, and then price elasticity for demand is calculated as a measure to quantify the impact of each factor. Furthermore, the first chapter provides elasticity values for the New York City and each of the five boroughs, and studies the relationship between these values and some socio-economic characteristics.
The second part of this dissertation studies the demand of taxi and how it is affected by other public transportation modes and weather. This demand modeling technique utilizes a combination of time series and linear regression models. The proposed method is then applied to yellow cab data in New York City. The pick-up points of yellow cab data in April, May, and June of 2014 are considered and aggregated every hour. The results show a significant correlation between taxi demand and demand for other transportation modes, as well as weather conditions. It is shown that combining time series models with linear regression will improve the performance of the model.
This study then follows by working on the time series models and considering the spatial variation of the demand. To understand the user demand for taxis through space and time, a generalized spatio-temporal autoregressive (STAR) model is proposed. In order to deal with the high dimensionality of the model, LASSO-type penalized methods are proposed to tackle the parameter estimation. The forecasting performance of the proposed models is measured using the out-of-sample mean squared prediction error (MSPE), and it is found that the proposed models outperform other alternative models such as vector autoregressive (VAR) models. The proposed modeling framework has an easily interpretable parameter structure and can feasibly be applied by taxi operators. The efficiency of the proposed model shows advantages for model estimation in real-time applications.
Furthermore, this dissertation studies the demand for e-hailing services which are growing rapidly especially in large cities. Similar to taxi demand, Uber demand is not distributed uniformly, either spatially or temporally, and this study proposes using spatio-temporal models to predict Uber demand as well. Moreover, the prediction performances of several statistical models are compared with each other: a) one temporal model (vector autoregressive (VAR)), b) two proposed spatio-temporal models (spatial-temporal autoregressive (STAR), c) least absolute shrinkage and selection operator applied on STAR (LASSO-STAR)). They are compared in different scenarios (based on the number of time and space lags), and for both peak and off-peak periods (rush hours and non-rush hours). This section additionally proposes different weighting matrices to improve the performance of the model. The results show the need to consider spatial models for e-hailing services and demonstrate significant improvement in the prediction of demand using the two proposed models.
Faghih, Sabiheh, "Understanding and Modeling Taxi Demand Using Time Series Models" (2019). CUNY Academic Works.