Real-time flight delay analysis and prediction based on the Internet of Things Data

Abstract

Flight delay is a significant problem resulting in the wasting of billions of dollars each year. Although this problem has been investigated in previous studies, all these previous studies rely on the historical records of flights provided by other agencies. Our work utilizes the emerging Internet of things (IoT) paradigm. It is now possible to collect and analyze sensors data in real-time. Our goal is to improve our understanding of the roots and signs of flight delays in order to be able to classify a given flight based on the features from flights and other data sources. We extend the existing works by adding new data sources and considering new factors in the analysis of flight delay. Through the use of real-time data, our goal is to establish a novel service to predict delays in real-time. In this project, we made a novel approach to collect the real time data from distributed sensors to study the flight delay. We create regression models to classify flights whether these flights are on-time or delayed as well as predicting how many minutes the delay would be. There are three main steps we conduct: first, we build a crawler to crawl the data from the pre-specified IoT data sources. Second, we implement an integration algorithm to integrate the data of all data sources using temporal and spatial criteria. Third, we conduct the analysis on the data with the aim to build a prediction model that could classify the flights and predict the delay time. This conducted analytical study provides three cases studies: Australia, China, and Europe. In addition, this project shows high correlation among the collected data. In addition, it shows that the prediction models in all case studies achieves very high accuracy. Comparing our models to others in previous studies, our model brings new factors that have impact on the flight delay as well as accomplish higher precision and recall.