Research For Aviation Data Mining And Knowledge Discovery Based On ASDI

The Aircraft Situation Display to Industry (ASDI) of the Traffic FlowManagement System (TFMS) of Federal Aviation Administrator (FAA) allows nearreal-time air traffic data to be accessible to members of the aviation industry. Theinformation can serve aviation related organizations as guidance for their decisionmaking and planning.This thesis aims to apply data mining theories and methods to find useful andpotential knowledge and patterns related to flight delay and critical air trafficregions from ASDI historical data, which in turn helps improving their business.Firstly it is principal to unify the format of ASDI data stored in xml files andtranslate them into relational database since ASDI data is stored with XML format,during which methods in xml processing and database management will be involved.In this process, based on the analysis of ASDI data, Entity-Relation model shall beextracted in order to ensure its consistency. Then the raw data shall be preprocessedthrough data classification, data reduction and data integration since the raw datacan be incomplete, noisy and with errors. Some key algorithms and methods such asregression analysis, data smoothing and clustering will be introduced and applied.When desired data is ready, data mining approaches can be achieved. The researchfocuses on two main aspects. These initial two steps are supposed to be the mostcomplex and difficult tasks. Finally, some data mining tools such as Weka will beapplied to enable knowledge discovery. Such tools apply theories and techniqueslike decision tree, cluster computation and rule prediction. Possible knowledge andpatterns can be found which can be valuable to decision and planning making ofaviation industry.