Non-linear algorithm analysis of large scale traffic data for systematic pattern detection

Large-scale sensor data of are well known to transportation researchers, but have resisted systematic analysis due to the significant challenges of dealing with missing data. Missing traffic data is a common problem in road traffic management when electronic devices do not work during some part of daily time. When some of traffic data has been lost, the continuity of traffic data cannot been achieved, so analyzing data will be face of lacking data. In this study, an artificial neural network (ANN) model has been choose to estimate the data, which have been lost in the Federal Highway Kuala Lumpur when Automatic Incident Detection (AID) devices had not been working during in the parts of daily time. Volume, vehicles speed, occupancy and vehicles gap are the major parameters will be consider for modeling. The present model will be evaluate with ANN. The NeuroXL predictor software will be use in this study. The prediction data will be compare with actual data for model performance comparison. Expected result from this study will show that ANN model successfully predict missing data based on some independent variables, which can be gather more easily.