Description
Date depot: 20 mars 2019
Titre: A robust flight price forecasting framework with uncertainty estimation
Directeur de thèse:
Pietro MICHIARDI (Eurecom)
Directeur de thèse:
Maurizio FILIPPONE (Eurecom)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
Forecasting plays a critical role within
the travel industry. The establishment of revenue management systems in the
eighties, led many actors to use data analysis and sophisticated mathematical
techniques in order to predict the willingness to pay of customers and maximise
their profits.
In the context of this PhD Thesis we
will explore statistical modelling techniques for the problem of long-term
forecasting of time series data. In particular, we will focus on challenges
related to: 1) algorithmic scalability, whereby the large scale nature of the
training data calls for distributed learning approaches; 2) robustness to noisy
data, such that the proposed algorithms will continue operation in spite of
potential problems in upstream methods of data collections; 3) accurate
quantification of uncertainty and calibrated models, such that the output of
the designed algorithms will be full predictive distributions, thus enabling
decision making with confidence information about the quality of predictions.
The Thesis is supported by Amadeus,
which will provide essential data, and real-world constraints to the scientific
problem addressed.
Doctorant.e: Candela Rosa