このエントリーをはてなブックマークに追加
ID 11427
Eprint ID
11427
フルテキストURL
タイトル(別表記)
Kernel density estimation on the interval
著者
植木 優夫 岡山大学
笛田 薫 岡山大学
抄録
In the field of data analysis, including environmental data, it is important to know the shape of underlying density function. In this case, we often use histogram which provides an information about the board line of density's curve. However histogram can not be the best method when the true density function is continuous, as is often the cases. On the other hand, kernel density estimator is another popular one which gives a continuous function. In some practical cases, however, there is a case that some knowledges about the range of the data are previously given. For instance, data of percentage, such as mortality rate, only takes the values on [0,1]. This paper considers two different modifications in kernel density estimator for the data on known interval and compares them.
キーワード
kernel density estimator
adaptive bandwidth
data on finite interval
発行日
2006-03-15
出版物タイトル
岡山大学環境理工学部研究報告
出版物タイトル(別表記)
Journal of the Faculty of Environmental Science and Technology, Okayama University
11巻
1号
出版者
岡山大学環境理工学部
出版者(別表記)
Faculty of Environmental Science and Technology, Okayama University
開始ページ
27
終了ページ
30
ISSN
1341-9099
NCID
AN10529213
資料タイプ
紀要論文
関連URL
http://www.okayama-u.ac.jp/user/est/homeJapan.html
言語
Japanese
論文のバージョン
publisher
査読
無し
Eprints Journal Name
fest