Car traffic is a fundamental problem of our modern world, and those who work on short term prediction of
traffic for navigation purposes as well as those who work on building long term solutions to car traffic
problems need accurately calibrated models to determine possible solutions. This work considers a new
way to build a relationship between car flow and density for use in a statistical model based on real world
data, as well as refines a fault detection algorithm using machine learning for identifying data sources that
are not accurate to the surrounding information.A version of this dissertation already exists, but is poorly
written, and | need assistance to improve the writing with a more narrative point.