Performances of Non-Parametric Statistics in Sensitivity Analysis and Parameter Ranking

A NUMBER OF MOSTLY NON-PAREAMETRIC SENSITIVITY ANALYSIS TECHNIQUES ARE COMPARED IN THE CASE OF NON-LINEAR MODEL RESPONSES. THE TEST MODELS ORIGINATE FROM THE CONTEXT OF THE RISK ANALYSIS FOR THE DISPOSAL OF RADIOACTIVE WASTE WHERE SENSITIVITY ANALYSIS PLAYS A CRUCIAL ROLE. ALL TECHNIQUES ARE APPLIED TO THE OUTPUT FROM THE SAME MONTE CARLO SIMUALTIONS WHERE A MODIFIED VERSION OF LATIN HYPERCUBE METHOD IS USED FOR THE SAMPLE SELECTION. HYPOTHESIS IS SYSTEMATICALLY APPLIED TO QUANTIFY THE DEGREE OF CONFIDENCE IN THE RESULTS GIVEN BY THE VARIOUS SENSITIVITY ESTIMATORS.NA-NOT AVAILABL