@article{JRC122315, number = {KJ-NA-30431-EN-N (online)}, address = {Luxembourg (Luxembourg)}, issn = {1831-9424 (online)}, year = {2020}, author = {Perrotta D and Checchi E and Torti F and Cerasa A and Arnes Novau X}, isbn = {978-92-76-24707-4 (online)}, publisher = {Publications Office of the European Union}, abstract = {The Customs Surveillance system (Surveillance Monitoring System (SMS) or Surveillance) of DG Taxation and Customs Union (DG TAXUD) centralises all European Union import and export declarations, collected from the national customs authorities on a daily basis according to Article 55(2) of Commission Implementing Regulation (EU)2015/2447. The Customs Surveillance data provide actual and prompt information about quantities, values, origin and destination of each traded commodity. The analysis of these trade flows poses a number of statistical challenges caused by the heterogeneous nature of the trade and the presence of many anomalous numerical values in the declarations (clerical errors, market peculiarities but also frauds). This report presents some solutions to these data analysis complications. The statistical approaches discussed in the report have been jointly developed by the Joint Research Centre and its academic partners, and have been applied in various broad application areas using our FSDA software library, based on the MATLAB environment. Their illustration is driven here by the specific needs of the Clearing House Task Force established by the European Commission to monitor the trade of COVID-19 related commodities during the pandemic which exploded in 2020. We show how this activity contributed to a relevant policy impact, by supporting the work for refining the definition of the codes used to trade protective face masks, adopted in the EU as of October 2020. In addition, the report is addressed to the services of the European Commission that are responsible for the Customs Surveillance system (DG TAXUD) and for its use in anti-fraud (European Anti-Fraud Office, OLAF). The Member States Customs authorities also benefit from these studies, because our methods and models are deployed in a customs anti-fraud resource jointly developed and maintained by the JRC and OLAF in the respective IT environments. }, title = {Addressing Price and Weight heterogeneity and Extreme Outliers in Surveillance Data}, url = {}, doi = {10.2760/817681 (online)} }