Imputation of Missing Traffic Flow Data Using Denoising Autoencoders

In transportation engineering, spatio-temporal data including traffic flow, speed, and occupancy are collected from different kinds of sensors and used by transportation engineers for analysis. However, the missing data influence the analysis and prediction results significantly. In this paper, Denoising Autoencoders are used to impute the missing traffic flow data. In our earlier research, we focused on a more general situation and used three kinds of Denoising Autoencoders: Vanilla, CNN, and Bi-LSTM, to impute the data with a general missing rate of 30%. The Autoencoder models are used to train on data with a high missing rate of about 80% in this paper. We demonstrate that even under extreme loss conditions, and Autoencoder models are very robust. By observing the hyper-parameter tuning process, the changing prediction accuracy is shown and in most cases, the three models maintain the original accuracy even under the worst situations. Moreover, the error patterns and trends concerning different sensor stations and different hours on weekdays and weekends are also visualized and analyzed. Finally, based on these results, we separate the data into weekdays and weekends, train and test the model respectively, and improve the accuracy of the imputation result significantly. (C) 2021 The Authors. Published by Elsevier B.V.

Automated summary

In this study, Denoising Autoencoders were used to fill in missing traffic flow data, improving prediction accuracy and demonstrating the robustness of the models under various conditions.