Dataset article 2

This is a follow up to the paper titled ”Modelling aircraft noise map around an airport using machine learning”. The objective of this study is to implement some additional parameters to the existing radar data, with the aim of improving the accuracy of the aircraft noise map by adding meteorologicalaerodrome reports (METAR). The noise map is built with the navigation function and the dataset for the machine learning is filled with merged information: radar real time data and METAR messages collected in the same period. However, the expected result of the fusion requires some deep analysis. In this study, we merge two information from two heterogeneous worlds into one information put inside an unique cartesian reference based on Bayesian method. The first technique to eliminate the gaps from METAR stations and the gray area of the radar is inverse distance weighted (IDW) for a ponderated interpolation and the second one is the space-time covariance function which quantifies the propagation of the phenomenas. As the machine learning efficiency and quality are based on the quality of the data, the cleaning data processing is controlling the fidelity and the synchronization of the merged information. The machine learning functions with three same algorithms such as Naive Bayes, Decision tree and Random forest.