Field evaluations and testing are critical to know the true performance of pavement markings, influenced by external environment and real world conditions. These installations and evaluations are a big commitment of time and resources demanding effective data collection. For durable marking materials with longer evaluation durations, the problems like missing inspections and censored values are often encountered, because of unexpected situations, weather constraints or due to actual work plan. While on the otherhand data points with regular intervals are often desired by statistical analysis and prediction methods. In this study, an imputation technique isdeveloped that can handle both the missing and censored points and can aid in designing a more economical alternative work plan. Retrore?ectivity data collected under the National Transportation Product Evaluation Program(NTPEP), from Mississippi test deck (2004) on U.S. Highway 78 is taken asa case study to show the application of imputation method, and there by forecasting using autoregressive and moving average (ARMA) models. Data analysis is carried out in two parts, method development (estimation) and application (validation and prediction). Imputation method is developed to estimate the missing values based on a class of Gaussian ARMA models. Finally ARMA models for time series data are used to validate and forecast the censored data points. Estimated missing values clearly emphasized the advantage of using imputation technique developed. Variations in the trend and order in the implementation of ARMA models have shown a clear improvement as the predicted measurements for the third year were reasonably accurate.With the developed imputation technique transportation agencies can effectively handle practical problems in the field evaluations and can design alternative economical work plans for field testing and evaluation of pavement marking materials.
Abstract