Sorry for my late notice. This week I am going to share some preliminary findings from an
ongoing paper “Classification Algorithms in Modeling Categorical Variables: A Comparison
of Multinomial Logistic Regression”. The paper focuses on addressing the two questions:
(1) if there any differences for algorithm performance under the conditions of relatively
large vs. small datasets, binary vs multiple, balanced vs unbalanced classes; (2) if there
any possibility to use algorithm learning to replace logistic statistics in hypothesis
testing for social science inquiry. Though few mathematical explanations will be provided,
I am looking forward to discussing with you about the answer to the above questions.