Travel time estimation of New York City taxis: An ensemble machine learning model

Travel Time Estimation (TTE) is a challenging problem that necessitates the use of advanced and adaptable Machine Learning (ML) models for accurate resolution. Its significance lies in the pervasive need for precise predictions of travel durations, as individuals increasingly rely on such information to efficiently manage their time in the context of their busy daily schedules. In this context, a novel ensemble machine learning model is proposed to address the TTE problem. The model is developed using the publicly available New York City taxi trip dataset, which serves as a comprehensive benchmark for building and evaluating ML models for TTE. To construct the proposed ensemble model, multiple ML algorithms are combined to leverage their complementary strengths. Additionally, Principal Component Analysis (PCA) and Truncated Singular Value Decomposition (TSVD) are individually employed as noise reduction techniques within the ensemble framework to enhance its robustness. The resulting ensemble models demonstrate superior performance, effectively improving overall prediction accuracy.