PL 49, A Method to Estimate Missing ATR Data

Abstract PL 49, A Method to Estimate Missing ATR Data Several recent projects have focused on estimating AADT for non ATR locations. A previous SPR study estimated ADT on locally maintained roads using a statistical modeling technique. A current OHS sponsored study uses existing data to build machine learning models with existing data (socio-demo, other) to estimate ADT for locally maintained roads with inventory data. A proposed OHS project will use ATR data to support traffic forecasting, e.g., better factors for a dashboard that can take new ATR data and make factors more accessible and frequent. In the current Planning Division-sponsored Streetlight evaluation project, we were provided with ATR counts by day for all Kentucky ATR locations. This data has many days with missing counts. For the current project, we addressed this issue by only considering ATR locations with seven or less missing counts. Thus, our results are only valid for ATRs with few missing values. Actual estimation of the missing values is the goal of this project.