Towards the construction of spatio-temporal data patterns in a road trafic urban network
par/by Marc Joliveau
Laboratoire de Mathématiques Appliquées aux Systèmes
École Centrale Paris
A wide variety of application domains have to deal with data providing from sensor networks. Unfortunately, the quality of such data is usually weak. Data sets are very noisy and often incomplete, due to several factors like partial system failures or bad conditions of measurements.
We will discuss about techniques to manipulate such data, concerning both dimensionality reduction and missing values, while keeping the inherent information. Then, it would be more easy to extract some knowledge and identify spatio-temporal correlation patterns.
In this way, we will introduce the so-called STPCA (Space Time Principal Component Analysis) method and illustrate its accuracy on a urban road traffic data set coming from a sensor network in a french big city.