Unbiased split selection for classification trees based on the Gini Index

By Carolin Strobl, Anne-Laure Boulesteix, Thomas Augustin. 2007

319 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Statistics and Probability, Statistical Learning Theory

How researchers see this paper

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

In Artificial Intelligence

In Information Systems

In Statistics and Probability

In Statistical Learning Theory

Citations in the corpus

Measuring Data (2022, methodology): “…psychology, this measure has been applied within Machine Learning for classiﬁcation decision trees to select data splits (Raileanu and Stoffel, 2004; Strobl et al., 2007), similar to the usage of KL-Divergence discussed in subsection 4.1, although it remains under-utilized in other approaches and…”
The in silico human surfaceome (2018, methodology): “Feature importance was assessed by using the Gini index (39) and revealed five features to be most important (Fig.”
Understanding Random Forests: From Theory to Practice (2014, methodology): “To guide our discussion, let us revisit the simulation studies from [Strobl et al., 2007b] which consider a binary output variable Y and five input variables X1, . . . ,X5, as defined in Table 7.1.”
A new variable importance measure for random forests with missing data (2012, methodology): “Likewise, predictors with many missing values may be preferred if the Gini Index is employed (Strobl et al. 2007).”
Applied Predictive Modeling (2013, methodology): “Also, as the number of missing values increases, the selection of predictors becomes more biased (Carolin et al. 2007). It is worth noting that the variable importance scores for the solubility regression tree (Fig. 8.6) show that the model tends to rely more on continuous (i.e., less granular) pred”
A Bias Correction Algorithm for the Gini Variable Importance Measure in Classification Trees (2008, methodology): “Another interesting approach is presented in [Strobl et al. (2007b)], where the exact distribution of the maximally selected Gini gain is derived by means of a combinatorial approach and the resulting p -value is suggested as an unbiased split selection criterion in recursive partitioning…”
Random Forest Algorithm for the Classification of Neuroimaging Data in Alzheimer's Disease: A Systematic Review (2017, methodology): “The Gini index is a measure of prediction power of variables in regression or classiﬁcation, based on the principle of impurity reduction (Strobl et al., 2007); it is non-parametric and therefore does not rely on data belonging to a particular type of distribution.”
Empirical characterization of random forest variable importance measures (2008, methodology): “In RF, P -value adjusted conditional inference CTs (Hothorn et al., 2006) together with subsampling without replacement should be used (Strobl et al., 2007a).”
Zeileis Conditional Variable Importance for Random Forests (2015, background): “However, it has been shown to be biased when predictor variables vary in their number of categories or scale of measurement [13], because the underlying Gini gain splitting criterion is a biased estimator and can be affected by multiple testing effects [14].”
Regression trees for longitudinal and multiresponse data (2012, background): “Because a categorical X with m unique values allows (2m−1 − 1) splits of the data and an ordinal X with n unique values allows (n− 1) splits, categorical variables with many unique values tend to have an advantage over ordinal variables in being selected [Loh and Shih (1997), Shih (2004), Strobl, Bo”