90 likes | 101 Views
Advanced toolkit for time series analysis with functionalities like outliers detection, forecasting, and seasonal adjustment. The toolkit is IT-friendly, portable, and supports various applications. Collaborative effort towards a common seasonal adjustment framework.
E N D
The NBB time series toolkit Towards a seasonal adjustment framework jean.palate@nbb.be
NBB TS toolkit. Goals • Functional • Providing solutions in the time series domain, for the production and the analysis of official statistics • Outliers detection, forecasting, missing values estimation, seasonal adjustment, benchmarking, temporal disaggregation, retropolation, cycle detection ... • IT • Fast and coherent developments, for a variety of contexts • from batch processing to advanced graphical interfaces • integration in commercial software or in in-house developments
NBB TS toolkit. Description • Based on standard technologies • [COM/C++], .NET, Java • Design • Fine granularity (> 500 classes) • From low-level entities to high-level modules. • No direct connection with specific data sources • Portable solution • Separate modules to access DBs
SA Framework. Context • Seasonal Adjustment Steering Group (SASG, ECB-Eurostat) • Guidelines and recommendations for the ESS. • IT sub-group of the SASG (9/2007) • Replacement of DEMETRA • Goals of the new software • Should reflect the recommandations of the SASG • Open (extensible, reusable), modular solution • Portable (not linked to a specific configuration) • Based on the current core engines (Tramo-Seats and X13) • Common modules (analysis, presentation, reports ...) for both algorithms • New architecture: similar to the NBB TS library
Final applications EXCEL ADD-INS "NEW DEMETRA" STAND-ALONE APPLICATIONS User interface toolkit NBB TS library SEASONAL ADJUSTMENT USER'S INTERFACES (PRESENTATION, FORMATTING) NBB, any other NSI, NCB, outsourcing DATABASES CONNECTIONS Any NSI, NCB, outsourcing High-level econometric libraries NBB TS library SEASONAL ADJUSTMENT MODULES (ANALYSIS, AUXILIARY ALGORITHMS) NBB, any other NSI, NCB, outsourcing CORE CONCEPTS NBB TS library TRAMO-SEATS (FORTRAN) BDE X13 (FORTRAN) USCB Low-level algorithms SA Framework. Architecture
SA Framework. Responsabilites • Specifications • Users' requirements, SA recommendations • SASG • Developments • Core engines • Bank of Spain, US Census Bureau (NBB for the COM interface) • Auxiliary modules • NBB (current TS library), other NSIs, NCBs ..., Outsourcing • Final applications(s) • Outsourcing • Organization, logistics: • Eurostat
Challenges • Stable core API (Object-model) • Testing (accuracy of the algorithms, reusability of the design) • Versioning • Documentation - Education - Users' community • Open source solution (?)
Concluding remarks • An IT solution based on reusable OO components is a valuable approach in the statistical domain (see NBB library). • The building of a common seasonal adjustment framework, using a similar approach and with the collaboration of several entities, is a challenging attempt.