Open AccessAn alternative approach in predictive modeling using model averaging scheme for logistic regression case (case study: application in class prediction of autistic spectrum disorder data)
Author(s) -
Septian Rahardiantoro,
Anang Kurnia,
Mokhamad Ramdhani Raharjo,
Yusma Yanti
Publication year - 2019
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/299/1/012039
Subject(s) - categorical variable , logistic regression , statistics , regression analysis , mathematics , model selection , logistic model tree , computer science , data mining
Logistic regression has become a popular method for handling predictive modeling when the response variable has a categorical scale. The difference in category proportion in response variable could influence the prediction accuracy. This research applied the model averaging approach for logistic regression in purpose to improve the prediction accuracy in different proportion of each category. Model averaging has the idea to combine some model candidates based on the specified weight to be the final model. The model candidate in model averaging generated based on all possibilities variable selection in the model. AIC weight is chosen to apply in the combination of all possible model candidates. It is illustrated with an application to data from a classification of Autistic Spectrum Disorder data. The result of this case indicated that the logistic model averaging had better performances.