The Multivariate k-Nearest Neighbor Model for Dependent Variables : One-Sided Estimation and Forecasting
This article gives the asymptotic properties of multivariate k-nearest neighbor regression estimators for dependent variables belonging to Rd, d > 1. The results derived here permit to provide consistent forecasts, and confidence intervals for time series An illustration of the method is given through the estimation of economic indicators used to compute the GDP with the bridge equations. An empirical forecast accuracy comparison is provided by comparing this non-parametric method with a parametric one based on ARIMA modelling that we consider as a benchmark because it is still often used in Central Banks to nowcast and forecast the GDP.
|Date of creation:||Dec 2009|
|Publication status:||Published in Documents de travail du Centre d'Economie de la Sorbonne 2009.50 - ISSN : 1955-611X. 2009|
|Note:||View the original document on HAL open archive server: https://halshs.archives-ouvertes.fr/halshs-00423871v2|
|Contact details of provider:|| Web page: https://hal.archives-ouvertes.fr/|
When requesting a correction, please mention this item's handle: RePEc:hal:cesptp:halshs-00423871. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (CCSD)
If references are entirely missing, you can add them using this form.