Travel Time Prediction Model For Nairobi City: An Application Of Machine Learning Algorithms

Abstract

Most studies done on mobility for Nairobi have sought to identify the causes of time variability in travel for the city but have not quantified the effect nor used them as the foundation of building prediction models. This research paper examines the application of machine learning algorithms in developing models that can predict travel times for Nairobi city based on historical taxi trip data made publicly available by Uber. A total of four datasets for the year 2018 covering weekday hourly travel data were used in developing the prediction models. three machine learning algorithms were used comprising of two ensemble learning methods and one normal standalone algorithms. The ensemble models were found to perform better prediction than the normal standalone model in terms of root mean squared error.