10. tailstat

tailstat: A Statistical Analysis Tool for the Long-Tail Era

This work shows that, in today's networked information society, there is growing importance in analyzing small samples for which the population distribution is known and the central limit theorem does not apply. It rebuilds statistical methods for that setting and develops a statistical analysis package

called tailstat that realizes them. Using a large-scale analysis of a social bookmarking site as a concrete case, it shows that analysis with tailstat is effective, and discusses how to apply it to other cases.

SourceForgeサイト：http://sourceforge.jp/projects/tailstat/
応用事例：tailstatによるデータマイニング事例：本棚.org版「僕らの大切な本」100選

Publications

栗原一貴, 土谷洋平: ロングテール時代のための中心極限定理によらない統計分析手法, 情報処理学会論文誌, Vol.52, No.2, (2011),pp.477-487 . PDF
Kazutaka Kurihara and Yohei Tsuchiya: "Development of tailstat, a Statistical Analysis Tool for the Long-Tail Era," IPSJ SIG Technical Report, Vol.2010-HCI-138, No.3, pp.1-8, 2010.
Kazutaka Kurihara, Yohei Tutiya, "Non-Central Limit Theorem Statistical Analysis for the "Long-tailed" Internet Society," http://arxiv.org/abs/1002.4682, 2010.