Accuracy Estimation and Comparison of Predictive Algorithms

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F62690094%3A18450%2F00%3A5538" target="_blank" >RIV/62690094:18450/00:5538 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Accuracy Estimation and Comparison of Predictive Algorithms

Popis výsledku v původním jazyce

Methods of the knowledge dicovery in databases allow assess information that is hidden in data. Results of data mining, the process of automating information discovery, can be used for predictive decisions. Component parts of predictive data mining are classification algorithms. Building the powerful model is impossible without comparing different algorithms. The use of classification accuracy as the only measure can be missleading in some circumstances. The results of ROC analysis and ROCCH are presented here for detection the dominating model (under different cost and clas distributions). The ROCCH is able to detect models that cannot be dominating under any circumstances. Logistic regression, discriminant analysis, and classification and regressiontress models are compared. Cross validation technique is used. Method ROCCH allowed to specify models that are superior for particular cost and class distribution.

Název v anglickém jazyce

Accuracy Estimation and Comparison of Predictive Algorithms

Popis výsledku anglicky

Methods of the knowledge dicovery in databases allow assess information that is hidden in data. Results of data mining, the process of automating information discovery, can be used for predictive decisions. Component parts of predictive data mining are classification algorithms. Building the powerful model is impossible without comparing different algorithms. The use of classification accuracy as the only measure can be missleading in some circumstances. The results of ROC analysis and ROCCH are presented here for detection the dominating model (under different cost and clas distributions). The ROCCH is able to detect models that cannot be dominating under any circumstances. Logistic regression, discriminant analysis, and classification and regressiontress models are compared. Cross validation technique is used. Method ROCCH allowed to specify models that are superior for particular cost and class distribution.

Druh

D - Stať ve sborníku

CEP obor

BB - Aplikovaná statistika, operační výzkum

OECD FORD obor

—

Projekt

—

Návaznosti

Z - Vyzkumny zamer (s odkazem do CEZ)

Rok uplatnění

2000

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proc. of the 18th Int. Conf. on Math. Methods in Economics

ISBN

80-245-0057-4

ISSN

—

e-ISSN

—

Počet stran výsledku

6

Strana od-do

—

Název nakladatele

Czech Society for Operations Research

Místo vydání

Praha

Místo konání akce

—

Datum konání akce

—

Typ akce podle státní příslušnosti

—

Kód UT WoS článku

—