TensorFlow implementation

TensorFlow

The TensorFlow Learn to Rank (TF-LTR) library was used to implement a machine learning technique to predict the rank order of landing airplanes. The goal is to develop an autonomous landing traffic management system. The TF-LTR library is equipped with techniques that will take features, a scoring function, evaluation metrics, and loss function and create a model that will accurately rank incoming data. This was applied to the ranking order scenario, with specific weather and ADS-B features chosen to help predict landing rank. After sufficient training of the model, this should result in a successful ranker that is able to process new testing data and correctly identify the true landing rank order of the airplanes.

Figure 1. TensorFlow Code FlowChart

This figure outlines the model building process.

Click this link to view the full code.