Tree Induction vs. Logistic Regression: A Learning-Curve Analysis

By Claudia Perlich, Foster J. Provost, Jeffrey S. Simonoff. Journal of machine learning research, 2003

392 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/70aefdc50ba5be828ce69aeb8303fb92ee8b17a1

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

Handling Missing Values when Applying Classification Models (2007, methodology): “…we use bagged classification trees (Breiman, 1996), which have been shown repeatedly to outperform simple classification trees consistently in terms of generalization performance (Bauer and Kohavi, 1999; Perlich et al., 2003), albeit at the cost of computation, model storage, and interpretability.”
An empirical comparison of supervised learning algorithms (2006, methodology): “Perlich et al. (2003) conducts an empirical comparison between decision trees and logistic regression.”
Machine learning and soil sciences: a review aided by machine learning tools (2019, background): “…depends on the complexity of the underlying problem, but the relationship between dataset size and performance has been shown in many studies, using what is usually known as “learning curves” (Catlett, 1991; Shavlik et al., 1991; Cortes et al., 1994; Perlich et al., 2003; Somarathna et al., 2017).”
Fifty Years of Classification and Regression Trees (2014, methodology): “M5 (Quinlan, 1992) and its implementation M50 (Wang & Witten, 1996) fit a piecewise multiple linear tree model by using a less exhaustive but much faster approach. M5 first grows a piecewise constant tree and then fits a stepwise multiple linear model in each node t , using as linear predictors only”
Homo Heuristicus: Why Biased Minds Make Better Inferences (2009, background): “The philosopher Rudolf Carnap (1947) proposed the ‘‘principle of total evidence,’’ the recommendation to use all the available evidence when estimating a probability.”
Financial Statement Fraud Detection: An Analysis of Statistical and Machine Learning Algorithms (2011, unknown): “However, the area under the curve for logistic regression (AUC = 0.823), the measure of signal separability used in Perlich et al. (2003), is between the low-and high-separability groups found in their study.”
OpenML: networked science in machine learning (2014, background): “It has been shown that studies regularly contradict each other because they are biased toward diﬀerent datasets [20], or because they don’t take into account the eﬀects of dataset size, parameter optimization and feature selection [33; 17].”
Integer optimization methods for machine learning (2012, methodology): “(2.5) To use logistic regression for ranking, we would use the probability estimates to rank the examples [e.g., Cooper et al., 1994, Fine et al., 1997, Perlich et al., 2003].”
Groundwater-Potential Mapping Using a Self-Learning Bayesian Network Model: A Comparison among Metaheuristic Algorithms (2021, methodology): “The cumulative distribution on → V can be calculated using Equation (3) [18,50].”
Predictive Modeling With Big Data: Is Bigger Really Better? (2013, unknown): “Optimal feature selection itself is a big data problem in such domains; further, using Naı¨ve Bayes we would not be able to determine a true upper bound on what would be possible using more sophisticated, non-linear modeling (cf., the striking differences in the learning curves for the linear and no”