Presenting logistic regression-based landslide susceptibility results
Citation
Lombardo, L., Mai, P.M. (2018). Presenting logistic regression-based landslide susceptibility results. Engineering Geology, 244: 14-24. Link to paper
Abstract
Although Logistic Regression models and methods have been widely used in geomorphology for several decades, no standards for presenting results in a consistent way have been adopted; most papers report parameters with different units and interpretations, therefore limiting potential meta-analytic applications. This paper proposes a standardization of Binary Logistic Regression analyses for landslide susceptibility and introduces a new work-flow to unify the way the community shares Logistic Regression results for landslide susceptibility purposes. A novel variable selection procedure is described and rescaled coefficients, variable importance, Jackknife tests, cutoff probability choice and response plots produce comparable results. The approach includes the Least Absolute Shrinkage Selection Operator (LASSO), a widely used approach in statistics for simultaneous parameter estimation and variable selection in generalized linear models. Simulations show that k-fold cross-validation with balanced landslide/no-landslide data affects probabilities and overestimate variance. The proposed workflow aims to improve reproducibility and comparability across landslide susceptibility studies using logistic regression.