A methodology to find the best classifier in business decision

 

Αποθηκεύτηκε σε:
Λεπτομέρειες βιβλιογραφικής εγγραφής
Συγγραφείς: Vega Vilca, José C., Torres Núñez, David A.
Μορφή: artículo original
Κατάσταση:Versión publicada
Ημερομηνία έκδοσης:2015
Περιγραφή:In this research, a methodology is presented to improve strategies of analysis in situations where supervised classification becomes the fundamental tool for business decision. The need to categorize the new customers into one of several groups, according to the characteristics of the subject, is analyzed through the calculation of the error rate. Programs were written using the statistical software package R, to calculate the error rate of each of nine classifiers, using cross-validation method 10 (Stone, 1974), in the 50 permutations of the data under consideration. For each of the analyzed data sets it was demonstrated, through ANOVA, that there are indeed significant differences in the average error rates of classifiers (p=0.00); therefore, it is concluded that the best classifier is the one with the lowest error rate.
Χώρα:Portal de Revistas UCR
Ίδρυμα:Universidad de Costa Rica
Repositorio:Portal de Revistas UCR
Γλώσσα:Español
OAI Identifier:oai:portal.ucr.ac.cr:article/19971
Διαθέσιμο Online:https://revistas.ucr.ac.cr/index.php/economicas/article/view/19971
Λέξη-Κλειδί :SUPERVISED CLASSIFICATION
CROSS VALIDATION
ERROR RATE
CUSTOMER
STATISTICAL DECISION
MULTIVARIATE ANALYSIS
CLASIFICACIÓN SUPERVISADA
VALIDACIÓN CRUZADA
TASA DE ERROR
CLIENTE
DECISIÓN ESTADÍSTICA
ANÁLISIS MULTIVARIABLE