Skip to main content
françaisFR
European Commission logo print header

AIR FOReCast in Europe

Objectif

The continuous increase of the number of air passengers in Europe (140 million passengers a year increasing by nearly 10% a year), together with the growing competition due to the easing of restrictions on trade is a permanent challenge for the airports and airline companies:
· Airports must schedule complex movement operations that are often distorted at the last moment by local traffic perturbation but also by cumulative delays coming from other parts of Europe. The emergence of the shuttle concept, where no reservation is necessary, further complicates the task,
· Airline companies have to organise logistic and maintenance operations, adjust their fleet and provide the best optimised carrier size for each destination, reserve slot allocations a very long time in advance according to expected passengers flows in Europe or in the world.
High performance tools or services able to predict the movements of passengers are required to help these companies to adjust their offer as closely as possible to the demand, or more simply to optimise their resources, would they be human or material. However, the complexity of these movements together with the random aspect of people's travel needs make it difficult to produce an adequate modelling based on reasoning or expertise.
Recent progress in data mining algorithms, if applied to this domain, could however allow to find correlations between causes and results in huge masses of events where a human being would simply be overwhelmed. Associated with an intelligent monitoring and interpretation of the information available on the Internet, these methods can be the seed of very innovative and efficient forecasting tools.

Objectives:
To provide better services to their customers or simply to optimise their resources, airports and airline companies need to forecast the number of air passengers, their destination and dates of travel. The factors motivating the European citizen to travel however are not easy to identify solely from human expertise because the number of potentially influencing parameters is too great. This project will evaluate the contribution of advanced statistical methods combining intelligent agents and data mining algorithms to forecast the number of air passengers for various destinations in Europe; develop methods allowing automatic extraction of limited sets of influencing parameters from masses of data; identify generic rules allowing extrapolation of results on departure destination couples for which little data is available; develop infrastructures capable of intelligent retrieval of useful data from existing data bases or the Internet in order to feed the forecast server developed during the project. An interesting output of the project will be a database of European events and trends, helping to identify what is happening and where, who travels, and when, which region is growing, etc.

Work description:
It is proposed to feed data mining algorithms to automatically extract relationships between events hidden in huge amounts of data and travel frequency.

The system will use 3 kinds of data: data from the past taken found in historical sources, data from the present found in reservation data bases, and data from the future obtained by monitoring of the Internet and automatic extraction of information in relation with travel incitement. This last point will require making a semantic analysis of the data collected to avoid overwhelming the system with irrelevant data. This will also require the analysis of several languages to assure a correct covering of the information sources. Up to four languages will be monitored: English, German, French and Italian, although the emphasis, due to resource limitations, will be on English. Finally, the extracted information will have to be transmitted to the data-mining engine under a meta-knowledge form, independent from the source language. The data mining algorithms will then allow detection of which subset of data has a real influence on the prevision.
A further step will consist in reducing the information space and be able to automatically sample among similar data which are sufficient to get good forecasts in order to limit the quantity of information submitted to the forecast system. Several prototypes devoted to specific problems of the end user will be developed to test the accuracy of the forecast. One of them will be implemented under the form of a web server, allowing distant consultation of travel flow forecasts and will pave the way to a further commercialisation of the technology to a large number of final users. The results will then be compared with methods or systems presently used by airline companies or airports during a final evaluation phase.

Milestones:
To start with, the project is working on dissemination plan and analysis of users requirements including market analysis, and will proceed with feasibility assessment and system detailed specifications with a first prototype with limited capabilities. Project will produce the report on information extraction progress and on prediction basis. Based on this information the second forecast prototype with full functionalities will be implemented in a client server architecture. The evaluation results on typical applications by end users of the Consortium will be reported.

Champ scientifique

Programme(s)

Thème(s)

Appel à propositions

Data not available

Régime de financement

CSC - Cost-sharing contracts

Coordinateur

SOFRESUD
Contribution de l’UE
Aucune donnée
Adresse
777 AVENUE DE BRUXELLES
83500 LA SEYNE SUR MER
France

Voir sur la carte
Coût total
Aucune donnée

Participants (6)

Partager cette page

Télécharger

Dernière mise à jour: 13 Juin 2005

Mon livret

Permalink: https://cordis.europa.eu/project/id/IST-1999-12179/fr

European Union, 2024