Analyse comparative entre les enquetes menages origine-destination et les systemes de paiement par carte a puce en transport urbain

Gathering quality information for the needs of transit planning has recently been preoccupying. Most of the actual sources of information based on a sample of the population, like the household travel survey, are encountering a reduction in the response rate; therefore it is harder to obtain a precise and acceptable sample.;Smart card collection fare system may be the answer to this issue. In fact, many transit network around the world are implementing smart card collection fare system or will implement them in a near future. The smart card can provide continuous day-by-day data of the trips made on the network. The data gathered by smart cards are unfortunately lacking information about the traveler and the purpose of the journey.;Data fusion is beginning to stand out as a potentially good solution to fill this need of information. The Societe de transport de I'Outaouais, a Canadian leader in innovative solutions for transit network, is already using a smart card collection fare system and have access to data from a 2005 household survey that took place in the area of Ottawa and Gatineau. This is a one of a kind opportunity to study those data and compare them since they were taken at the exact same time.;The primary goal of this work is to compare two different databases coming from two distinctive sources (a household survey and smart cards system). The studies are based on an object-oriented approach and they were divided into three categories, the global analysis, the segmented analysis and the network analysis. The global analysis showed a considerably large gap between the data from the household survey and the smart cards. But, those analyses are limited because they are aggregated.;The data used in the segmented analyses are disaggregated and thus permit to study objects like the days, the hours and the monthly pass holders. Those analyses showed that the discrepancy between the two datasets is local instead of global. Also, a study of the day-by-day variability of travels is done and it gives us a better understanding of the user behaviors on that particular transit network.;Network analyses were made with the objects routes and bus stops and, like the segmented analyses, they use disaggregated data. A study of the temporal variability on particular routes showed that the students and the adults have similar behaviors.;Analyzing origin-destination matrices confirmed that the estimates from the smart cards are more accurate when it comes to short-distance travels. In facts, analyzing the distribution of the trips, from both databases, by their origin and destination shows lower trip rate from the travel household survey.;After all those comparisons, it has been shown that a bias exists in both datasets and a corrective method has been applied to the household survey estimates. This method showed promising and encouraging results for further research and it demonstrated that it is possible to modify or calculate household estimates from other sources like smart cards database.;A discussion about data fusion techniques confirmed the lack of experiment in the field of knowledge of transit transportation and proposed a terminology of different methods of fusion. Some improvements for the two methods analyzed in this work were proposed to facilitate the integration of data. Finally, thoughts on the future of data fusion and on methods to spread errors codes across the databases showed the progress that has been made and what still need to be done.