Presenting logistic regression-based landslide susceptibility results

• Published in: Engineering geology 2018-10, Vol.244, p.14-24
• Main Authors: Lombardo, Luigi, Mai, P. Martin
• Format: Article
• Language: English
• Subjects: Binary logistic regression; Landslide susceptibility; Least Absolute Shrinkage Selection Operator (LASSO); Standardized results
• ISSN: 0013-7952; 1872-6917
• DOI: 10.1016/j.enggeo.2018.07.019

A new work-flow is proposed to unify the way the community shares Logistic Regression results for landslide susceptibility purposes. 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. We first summarize the major differences in the geomorphological literature and then investigate each one proposing current best practices and few methodological developments. The latter is mainly represented by a widely used approach in statistics for simultaneous parameter estimation and variable selection in generalized linear models, namely the Least Absolute Shrinkage Selection Operator (LASSO). The North-easternmost sector of Sicily (Italy) is chosen as a straightforward example with well exposed debris flows induced by extreme rainfall. •A standardization of Binary Logistic Regression analyses for landslide susceptibility is proposed.•A novel variable selection procedure is described.•Rescaled coefficients, variable importance, Jackknife tests, cutoff probability choice and response plots produce comparable results.•Simulations show that k-fold cross-validation with balanced landslide/no-landslide data affects probabilities and overestimate variance.