Multiple hypothesis testing in data mining

Sami Hanhijärvi

Multiple hypothesis testing in data mining

Julkaisun otsikon käännös: Monen hypoteesin testaus tiedonlouhinnassa

Sami Hanhijärvi

Tutkimustuotos: Doctoral Thesis › Collection of Articles

Abstrakti

Tiedonlouhinnan menetelmillä pyritään löytämään annetusta aineistosta yllättäviä ja mielenkiintoisia säännönmukaisuuksia, joita kutsutaan hahmoiksi. Useat menetelmät kuitenkin löytävät hahmoja kaikista aineistoista, jopa täysin satunnaisista. Näissä tilanteissa voidaan käyttää tilastollista testausta valitsemaan yllättävät hahmot, jotka eivät esiinny yhtä vahvasti satunnaisessa aineistossa. Monen hahmon tilastollista merkittävyyttä testatessa käsitellään samalla yhdenaikaisesti joukkoa tilastollisia hypoteesejä. Usean hypoteesin yhdenaikaista testausta kutsutaan monen hypoteesin testaamiseksi, joka vaatii erityistoimenpiteitä, jotta väärien johtopäätösten todennäköisyyttä voidaan hallita. Kuitenkaan tyypillisiä monen hypoteesin testausmenetelmiä ei voida käyttää tiedonlouhinnassa, koska ne eivät ota huomioon tiedonlouhinnassa tyypillistä vaihtelevan hypoteesijoukon ongelmaa. Tämä väitöskirja esittelee ongelman ja tarkastelee aiheeseen liittyviä julkaisuja. Kirja keskittyy monen hypoteesin testaamiseen erityisesti tiedonlouhinnan tilanteissa. Tyypillisten monen hypoteesin testaamiseen käytettävien menetelmien ongelmia tiedonlouhinnassa käsitellään, ja ongelmiin esitetään ratkaisu. Tämä perustuu satunnaistukseen, jossa luodaan satunnaisia aineistoja ja käytetään tiedonlouhinnan menetelmää näihin aineistoihin. Saatuja tuloksia verrataan alkuperäisestä aineistosta saatuihin tuloksiin. Satunnaistaminen esitellään yleisesti ja käsitellään mahdollisia satunnaistamismenetelmiä erilaisissa tiedonlouhinnan tilanteissa. Esitettyä ratkaisua käytetään iteratiivisessa tiedonlouhinnassa ja kaksoisryhmittelyssä, joissa kokeellisesti myös osoitetaan ratkaisun hyöty.

Julkaisun otsikon käännös	Monen hypoteesin testaus tiedonlouhinnassa
Alkuperäiskieli	Englanti
Pätevyys	Tohtorintutkinto
Myöntävä instituutio	Aalto-yliopisto
Valvoja/neuvonantaja	Mannila, Heikki, Vastuuprofessori Rousu, Juho, Vastuuprofessori Puolamäki, Kai, Ohjaaja
Kustantaja	Aalto University
Painoksen ISBN	978-952-60-4604-4
Sähköinen ISBN	978-952-60-4605-1
Tila	Julkaistu - 2012
OKM-julkaisutyyppi	G5 Artikkeliväitöskirja

Tutkimusalat

tiedonlouhinta
monen hypoteesin testaus
tilastollinen testaus
kaksoisryhmittely

Pääsy asiakirjaan

http://urn.fi/URN:ISBN:978-952-60-4605-1

Siteeraa tätä

@phdthesis{b31b2390d4e34eda9317f541f841da9f,

title = "Multiple hypothesis testing in data mining",

abstract = "Data mining methods seek to discover unexpected and interesting regularities, called patterns, in presented data sets. However, the methods often return a collection of patterns for any data set, even a random one. Statistical significance testing can be applied in these scenarios to select the surprising patterns that do not appear as clearly in random data. As each pattern is tested for significance, a set of statistical hypotheses are considered simultaneously. The multiple comparison of several hypotheses simultaneously is called multiple hypothesis testing, and special treatment is required to adequately control the probability of falsely declaring a pattern statistically significant. However, the traditional methods for multiple hypothesis testing can not be used in data mining scenarios, because these methods do not consider the problem of varying set of hypotheses, which is inherent in data mining. This thesis provides an introduction to the problem and reviews some published work on the subject. The focus is in multiple hypothesis testing and specifically in data mining. The problems with traditional multiple hypothesis testing methods in data mining scenarios are discussed, and a solution to these problems is presented. The solution uses randomization, which involves drawing samples of random data sets and using the data mining algorithm with them. The results on the random data sets are then compared with the results on the original data set. Randomization is introduced and discussed in general, and possible randomization schemes in different data mining scenarios are presented. The solution is applied in iterative data mining and biclustering scenarios. Experiments are carried out to display the utility in these applications.",

keywords = "data mining, multiple hypothesis testing, statistical significance testing, biclustering, tiedonlouhinta, monen hypoteesin testaus, tilastollinen testaus, kaksoisryhmittely, data mining, multiple hypothesis testing, statistical significance testing, biclustering",

author = "Sami Hanhij{\"a}rvi",

year = "2012",

language = "English",

isbn = "978-952-60-4604-4",

series = "Aalto University publication series DOCTORAL DISSERTATIONS",

publisher = "Aalto University",

number = "51",

address = "Finland",

school = "Aalto University",

}

TY - BOOK

T1 - Multiple hypothesis testing in data mining

AU - Hanhijärvi, Sami

PY - 2012

Y1 - 2012

N2 - Data mining methods seek to discover unexpected and interesting regularities, called patterns, in presented data sets. However, the methods often return a collection of patterns for any data set, even a random one. Statistical significance testing can be applied in these scenarios to select the surprising patterns that do not appear as clearly in random data. As each pattern is tested for significance, a set of statistical hypotheses are considered simultaneously. The multiple comparison of several hypotheses simultaneously is called multiple hypothesis testing, and special treatment is required to adequately control the probability of falsely declaring a pattern statistically significant. However, the traditional methods for multiple hypothesis testing can not be used in data mining scenarios, because these methods do not consider the problem of varying set of hypotheses, which is inherent in data mining. This thesis provides an introduction to the problem and reviews some published work on the subject. The focus is in multiple hypothesis testing and specifically in data mining. The problems with traditional multiple hypothesis testing methods in data mining scenarios are discussed, and a solution to these problems is presented. The solution uses randomization, which involves drawing samples of random data sets and using the data mining algorithm with them. The results on the random data sets are then compared with the results on the original data set. Randomization is introduced and discussed in general, and possible randomization schemes in different data mining scenarios are presented. The solution is applied in iterative data mining and biclustering scenarios. Experiments are carried out to display the utility in these applications.

AB - Data mining methods seek to discover unexpected and interesting regularities, called patterns, in presented data sets. However, the methods often return a collection of patterns for any data set, even a random one. Statistical significance testing can be applied in these scenarios to select the surprising patterns that do not appear as clearly in random data. As each pattern is tested for significance, a set of statistical hypotheses are considered simultaneously. The multiple comparison of several hypotheses simultaneously is called multiple hypothesis testing, and special treatment is required to adequately control the probability of falsely declaring a pattern statistically significant. However, the traditional methods for multiple hypothesis testing can not be used in data mining scenarios, because these methods do not consider the problem of varying set of hypotheses, which is inherent in data mining. This thesis provides an introduction to the problem and reviews some published work on the subject. The focus is in multiple hypothesis testing and specifically in data mining. The problems with traditional multiple hypothesis testing methods in data mining scenarios are discussed, and a solution to these problems is presented. The solution uses randomization, which involves drawing samples of random data sets and using the data mining algorithm with them. The results on the random data sets are then compared with the results on the original data set. Randomization is introduced and discussed in general, and possible randomization schemes in different data mining scenarios are presented. The solution is applied in iterative data mining and biclustering scenarios. Experiments are carried out to display the utility in these applications.

KW - data mining

KW - multiple hypothesis testing

KW - statistical significance testing

KW - biclustering

KW - tiedonlouhinta

KW - monen hypoteesin testaus

KW - tilastollinen testaus

KW - kaksoisryhmittely

KW - data mining

KW - multiple hypothesis testing

KW - statistical significance testing

KW - biclustering

M3 - Doctoral Thesis

SN - 978-952-60-4604-4

T3 - Aalto University publication series DOCTORAL DISSERTATIONS

PB - Aalto University

ER -

Multiple hypothesis testing in data mining

Abstrakti

Tutkimusalat

Pääsy asiakirjaan

Sormenjälki

Siteeraa tätä