Flexible and Robust Method for Missing Loop Detector Data Imputation

1 This work is primarily focused on missing traffic sensor data imputation for the purpose of improving the 2 coverage and accuracy of traffic analysis and performance estimation. Missing data, whether attributable 3 to hardware failure or error detection and removal, is a constant problem in loop and other traffic detector 4 datasets. As the rate of missingness increases, the treatment of missing values quickly becomes the 5 controlling factor in overall data quality. Previously, a number of imputation approaches have been 6 developed for traffic data. However, few studies aim at handling the traffic data with large blocks of missing 7 values for network-wide implementation. A proven predictive mean matching multiple imputation method 8 is introduced and applied to loop detector volume data collected on Interstate 5 in Washington State. Using 9 the iterative multiple imputation by chained equations approach, the spatial correlation between nearby 10 detectors is considered for prediction and the presence of missing data in all predictors is effectively dealt 11 with. The proposed methodology is shown to perform well on a range of missing data patterns including 12 missing completely at random, missing days, and missing months. After applying the imputation method 13 to 20-second data and performing post-imputation aggregation, the results in this study suggest that the 14 proposed method can outperform elementary pairwise regression and produce reliable imputation estimates, 15 even when entire days and months are missing from the dataset. Thus, the predictive mean matching 16 multiple imputation method can be used as an effective approach for imputing missing traffic data in a 17 range of challenging scenarios. 18 K. C. Henrickson, Y. Zou, and Y. Wang 3