eprintid: 227 rev_number: 20 eprint_status: archive userid: 2 dir: disk0/00/00/02/27 datestamp: 2015-02-20 11:30:32 lastmod: 2020-03-24 08:44:17 status_changed: 2015-04-27 08:05:58 type: article metadata_visibility: show creators_name: Hart, Jeffrey creators_name: Vieu, Philippe corp_creators: Texas A&M University corp_creators: Iniversite Paul Sabatier title: Data-Driven Bandwidth Choice for Density Estimation Based on Dependent Data ispublished: pub subjects: MP2 divisions: EPOS-P full_text_status: none abstract: The bandwidth selection problem in kernel density estimation is investigated in situations where the observed data are dependent. The classical leave-out technique is extended, and thereby a class of cross-validated bandwidths is defined. These bandwidths are shown to be asymptotically optimal under a strong mixing condition. The leave-one out, or ordinary, form of cross-validation remains asymptotically optimal under the dependence model considered. However, a simulation study shows that when the data are strongly enough correlated, the ordinary version of cross-validation can be improved upon in finite-sized samples. date: 1990 publication: The Annals of Statistics volume: 18 number: 2 publisher: Institute of Mathematical Statistics pagerange: 873-890 id_number: doi:10.1214/aos/1176347630 refereed: TRUE issn: 0090-5364 official_url: http://dx.doi.org/10.1214/aos/1176347630 access_IS-EPOS: unlimited owner: Publisher citation: Hart, Jeffrey and Vieu, Philippe (1990) Data-Driven Bandwidth Choice for Density Estimation Based on Dependent Data. The Annals of Statistics, 18 (2). pp. 873-890. DOI: https://doi.org/10.1214/aos/1176347630