Comparative Analysis of Machine Learning Models for Predicting Travel Time

TL;DR

This study compares five deep learning and statistical models for travel time prediction using Missouri data, finding ARIMA to be the most effective architecture for forecasting travel times.

Contribution

It provides a comparative analysis of multiple models, including ARIMA, RNN, AR, LSTM, and GRU, for travel time prediction, highlighting ARIMA's superior performance.

Findings

ARIMA outperformed other models in accuracy.

Optimal learning rate identified as 0.001.

Deep learning models showed competitive results.

Abstract

In this paper, five different deep learning models are being compared for predicting travel time. These models are autoregressive integrated moving average (ARIMA) model, recurrent neural network (RNN) model, autoregressive (AR) model, Long-short term memory (LSTM) model, and gated recurrent units (GRU) model. The aim of this study is to investigate the performance of each developed model for forecasting travel time. The dataset used in this paper consists of travel time and travel speed information from the state of Missouri. The learning rate used for building each model was varied from 0.0001-0.01. The best learning rate was found to be 0.001. The study concluded that the ARIMA model was the best model architecture for travel time prediction and forecasting.