Master'sOpen Access

Different estimation methods in multinomial logistic regression

Is this your thesis?

This record came from a bulk archive import. If it’s yours, link it to your profile.

2024
0 views
0 downloads

Abstract (EN)

Logistic regression analysis is a type of regression in which a categorical response variable is estimated with the help of one or more explanatory variables. Explanatory variables can be continuous, discrete, or a mixture of these. Logistic regression analysis is frequently used in classification problems as a supervised machine learning method. Binary (two-category) logistic regression is applied in data sets where the response variable has two categories, and multinomial (multi-category) logistic regression analysis is applied in data with three or more categories. In this study, multinomial logistic regression analysis performed on four different real data sets, all explanatory variables of which are continuous. The correlations between the explanatory variables were checked and multicollinearity was examined. The results were established for all data sets by Maximum Likelihood, Ridge, LASSO, Elastic-net and Principal Component Analysis methods. The prediction accuracy rates of the models in the test data sets were compared. In addition, the performances of five different models were examined by reducing the N/p ratio in a controlled manner (N: Number of observations, p: Number of explanatory variables) with randomly taken samples from each data set.

Author

Ekrem Bozbey

How to Cite

Ekrem Bozbey (Master Thesis). Different estimation methods in multinomial logistic regression, 2024, Tekirdağ Namık Kemal University.

Keywords

License

Tüm Hakları Saklıdır

This work is shared under the specified license terms.

More theses from Tekirdağ Namık Kemal University