A New Robust Approach for Multinomial Logistic Regression With Complex Design Model

A new estimation approach based on phi‐divergence measures for one‐shot device accelerated life testing

“…Thus, we may find more differences of behavior for small sample sizes. See 22 for a more detailed explanation on these asymptotic properties.…”

“…See refs. [ 8,[21][22][23][24] ], among many others. The phidivergence between (11) and ( 12) is given by…”

“…See refs. [ 8,21–24 ], among many others. The phi‐divergence between () and () is given by $$$$\begin{align*} d_{\phi }{\left(\widehat{\bm{p}},\bm{p}(\bm{\theta }) \right)}=\sum _{i=1}^{I}\sum _{j=1}^{2} p_{ij}(\bm{\theta })\phi {\left(\frac{\widehat{p}_{ij}}{p_{ij}(\bm{\theta })} \right)}, \end{align*}$$$$where $$\phi \in \Phi ^{\ast }$$ and $$\Phi ^{\ast }$$ denotes the class of all convex functions $$\phi:[0,\infty)\rightarrow \mathbb {R}\cup \lbrace \infty \rbrace$$ such that $$\phi (1) =0$$, $$\phi ^{\prime \prime }(1) >0$$ and we define $$0\phi (0/0) =0$$ and $$0\phi (p/0)=p\lim _{u\rightarrow \infty }\phi (u)/u$$…”

Subgroup State Prediction under Different Noise Levels Using MODWT and XGBoost

“…In [11] logistic regression models were used in genomic studies to analyze the genetic data linked to electronic health records, and their performance in the presence of positive errors in event time was evaluated. Castilla and Chocano [12] developed robust estimators and Wald-type tests for the multinomial logistic regression based on ϕ 2…”

“…In [ 11 ] logistic regression models were used in genomic studies to analyze the genetic data linked to electronic health records, and their performance in the presence of positive errors in event time was evaluated. Castilla and Chocano [ 12 ] developed robust estimators and Wald-type tests for the multinomial logistic regression based on ϕ -divergence measures and analyzed the robustness of the approach. Dumitrescu et al [ 13 ] proposed a credit-scoring model based on an adaptive LASSO logistic regression model with predictors extracted from decision trees, which give rise to a significant reduction in misclassification costs compared to the benchmark logistic regression.…”