All Times ET
Keywords: Unbalanced Data; MCMC; Neural Networks, Artificial Intelligence; Machine Learning; Logistic Re- gression; Categorical Data Analysis; Bayesian Estima- tion; Model Fit; Classification; Inference.
This article introduces a nonparametric methodology for Generalized Linear Models (GLMs), through an effective transformation of the current framework. It is shown to be an extension of the recent parametric contribution giving results superior to it in various settings. Despite being nonparametric we show that the methodology does not need more iterations in general for convergence in comparison to the parametric version if the underlying DGP is symmetric. If the underlying DGP is asymmetric it gives uniformly better prediction and inference performance to existing methodologies compared. Furthermore, we present a new classification statistics utilizing which we show that overall it has better inference and classification performance than the parametric version, which is statistically significant especially if the DGP asymmetric. We further show that the methodology can outperform existing Artificial Intelligence and Machine Learning methods such as Neural Networks. In addition, we show that the methodology can be used to perform model diagnostics for any categorical model, to check whether the data indeed comes from some parametric distribution of choice under minimal regularity conditions. This is a highly useful result which extends the work of Liu and Zhang (2018), to any GLM and is entirely novel to the sciences. Finally, we apply the methodology to various real-world datasets to compare and contrast with existing results to show where and how the new proposed methodology can outperform the other methods considered.