Sélection of subsets of regression variables

By Alan J. Miller. 1984

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Interpretable classification models for recidivism prediction (2015, background): “the inﬂuence of one input variable with respect to the others, which is helpful because most humans are seriously limited in the number of cognitive entities they can handle at once (7 2 according to Miller, 1984), and seriously limited in estimating the association between three or more variables (”
Learning Optimized Risk Scores (2016, background): “integer coefcients let users make quick predictions, without a computer or a calculator. In light of research showing that humans are limited in handling multiple cognitive entities (7 2 according to Miller, 1984) and in estimating the association between three or more variables (Jennings et al., 19”
Supersparse linear integer models for optimized medical scoring systems (2015, background): “Further, sparsity and small integer coefficients are also important because humans can only handle a few cognitive entities at once (7± 2 according to Miller, 1984), and are seriously limited in estimating the association between three or more variables (Jennings et al., 1982).”
Methods and Models for Interpretable Linear Classification (2014, background): “Sparsity : According to Miller [80], humans can only handle a few cognitive entities at once (7± 2).”
Recent advances in feature selection and its applications (2017, background): “Since the research in feature selection has been a challenging field, some researchers have doubted about its computational feasibility, such as in the paper [4].”
Selection of Variables for Cluster Analysis and Classification Rules (2006, methodology): “g variable, such that the new subset maximizes the number of successfully data allocations. We also consider replacement of previously introduced variables following the iterative scheme described by Miller (1984)as avariationofthe classical forward-backward methods. The increment continues until th”
A more accurate method for colocalisation analysis allowing for multiple causal variants (2021, background): “The stepwise regression approach upon which conditioning is based is known generally to produce potentially unreliable results[3], a phenomenon that can be exacerbated by the extensive correlation between genetic variants caused by linkage disequilibrium (LD)[4].”
Stepwise Model Fitting and Statistical Inference: Turning Noise into Signal Pollution (2008, background): “Consequently, the final model is not generally the best model (Miller 1984).”
Statistical Analysis in Climate Research (2000, unknown): “Ross [332] citing Copas [91] and Miller [278] points out that using the same sample to select the model and estimate its coefficients is ‘overfitting’ and can lead to models that perform very poorly on independent data.”
Recent advances and emerging challenges of feature selection in the context of big data (2015, background): “Miller’s paper’’ [10], for example, RL Plackett stated: ‘‘If variable elimination has not been sorted out after two decades of work assisted by high-speed computing, then perhaps the time has come to move on to other problems’’.”
Variable Selection Using Random Forests (2006, methodology): “Many automatic variable selection techniques have been proposed in the literature, for example the backward or forward stepwise regression (see Miller (1984) and Hocking (1976)) or the recent stepwise bootstrap method of Austin and Tu (2004).”
Supersparse Linear Integer Models for Interpretable Classification (2013, background): “Given that interpretability is an inherently multifaceted notion, SLIM is built to train models that are: – Sparse : It is well-known that humans can only handle a few cognitive entities at once (7 ± 2 according to [51]).”