scholarly journals Multidimensional Sensitivity Analysis of Large-Scale Mathematical Models

2013 ◽  
pp. 137-156 ◽  
Author(s):  
Ivan Dimov ◽  
Rayna Georgieva
Keyword(s):  
Sensitivity Analysis ◽  
Mathematical Models ◽  
Large Scale

scholarly journals The effects of flooding and weather conditions on leptospirosis transmission in Thailand

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Sudarat Chadsuthi ◽  
Karine Chalvet-Monfray ◽  
Anuwat Wiratsudakul ◽  
Charin Modchang
Keyword(s):  
Sensitivity Analysis ◽  
Public Education ◽  
Mathematical Models ◽  
Disease Transmission ◽  
Transmission Rate ◽  
Weather Conditions ◽  
Transmission Dynamics ◽  
Multiplication Rate ◽  
High Degree ◽  
Transmission Models

AbstractThe epidemic of leptospirosis in humans occurs annually in Thailand. In this study, we have developed mathematical models to investigate transmission dynamics between humans, animals, and a contaminated environment. We compared different leptospire transmission models involving flooding and weather conditions, shedding and multiplication rate in a contaminated environment. We found that the model in which the transmission rate depends on both flooding and temperature, best-fits the reported human data on leptospirosis in Thailand. Our results indicate that flooding strongly contributes to disease transmission, where a high degree of flooding leads to a higher number of infected individuals. Sensitivity analysis showed that the transmission rate of leptospires from a contaminated environment was the most important parameter for the total number of human cases. Our results suggest that public education should target people who work in contaminated environments to prevent Leptospira infections.

Parameter Sensitivity Analysis of a Large-Scale System With Several Components Interacting in Series

1989 ◽  
Author(s):  
H. Torab
Keyword(s):  
Sensitivity Analysis ◽  
Large Scale ◽  
Optimization Problems ◽  
Parameter Sensitivity ◽  
Engineering Optimization ◽  
Parameter Sensitivity Analysis ◽  
Large Scale Systems ◽  
In Series ◽  
Special Case ◽  
Engineering Optimization Problems

Abstract Parameter sensitivity for large-scale systems that include several components which interface in series is presented. Large-scale systems can be divided into components or sub-systems to avoid excessive calculations in determining their optimum design. Model Coordination Method of Decomposition (MCMD) is one of the most commonly used methods to solve large-scale engineering optimization problems. In the Model Coordination Method of Decomposition, the vector of coordinating variables can be partitioned into two sub-vectors for systems with several components interacting in series. The first sub-vector consists of those variables that are common among all or most of the elements. The other sub-vector consists of those variables that are common between only two components that are in series. This study focuses on a parameter sensitivity analysis for this special case using MCMD.

scholarly journals Machine learning-based feature importance approach for sensitivity analysis of steel frames

2021 ◽  
Author(s):  
Hyeyoung Koh ◽  
Hannah Beth Blum
Keyword(s):  
Machine Learning ◽  
Sensitivity Analysis ◽  
Feature Selection ◽  
Large Scale ◽  
Failure Modes ◽  
Model Development ◽  
Predictive Performance ◽  
Computational Effort ◽  
Structural Systems ◽  
Feature Importance

This study presents a machine learning-based approach for sensitivity analysis to examine how parameters affect a given structural response while accounting for uncertainty. Reliability-based sensitivity analysis involves repeated evaluations of the performance function incorporating uncertainties to estimate the influence of a model parameter, which can lead to prohibitive computational costs. This challenge is exacerbated for large-scale engineering problems which often carry a large quantity of uncertain parameters. The proposed approach is based on feature selection algorithms that rank feature importance and remove redundant predictors during model development which improve model generality and training performance by focusing only on the significant features. The approach allows performing sensitivity analysis of structural systems by providing feature rankings with reduced computational effort. The proposed approach is demonstrated with two designs of a two-bay, two-story planar steel frame with different failure modes: inelastic instability of a single member and progressive yielding. The feature variables in the data are uncertainties including material yield strength, Young’s modulus, frame sway imperfection, and residual stress. The Monte Carlo sampling method is utilized to generate random realizations of the frames from published distributions of the feature parameters, and the response variable is the frame ultimate strength obtained from finite element analyses. Decision trees are trained to identify important features. Feature rankings are derived by four feature selection techniques including impurity-based, permutation, SHAP, and Spearman's correlation. Predictive performance of the model including the important features are discussed using the evaluation metric for imbalanced datasets, Matthews correlation coefficient. Finally, the results are compared with those from reliability-based sensitivity analysis on the same example frames to show the validity of the feature selection approach. As the proposed machine learning-based approach produces the same results as the reliability-based sensitivity analysis with improved computational efficiency and accuracy, it could be extended to other structural systems.

scholarly journals Large-scale groundwater modeling using global datasets: a test case for the Rhine-Meuse basin

2011 ◽  
Vol 8 (2) ◽  
pp. 2555-2608 ◽  
Author(s):  
E. H. Sutanudjaja ◽  
L. P. H. van Beek ◽  
S. M. de Jong ◽  
F. C. van Geer ◽  
M. F. P. Bierkens
Keyword(s):  
Sensitivity Analysis ◽  
Land Surface ◽  
Large Scale ◽  
Groundwater Modeling ◽  
Land Surface Model ◽  
Water Levels ◽  
Surface Model ◽  
Groundwater Model ◽  
Model Output ◽  
Global Datasets

Abstract. Large-scale groundwater models involving aquifers and basins of multiple countries are still rare due to a lack of hydrogeological data which are usually only available in developed countries. In this study, we propose a novel approach to construct large-scale groundwater models by using global datasets that are readily available. As the test-bed, we use the combined Rhine-Meuse basin that contains groundwater head data used to verify the model output. We start by building a distributed land surface model (30 arc-second resolution) to estimate groundwater recharge and river discharge. Subsequently, a MODFLOW transient groundwater model is built and forced by the recharge and surface water levels calculated by the land surface model. Although the method that we used to couple the land surface and MODFLOW groundwater model is considered as an offline-coupling procedure (i.e. the simulations of both models were performed separately), results are promising. The simulated river discharges compare well to the observations. Moreover, based on our sensitivity analysis, in which we run several groundwater model scenarios with various hydrogeological parameter settings, we observe that the model can reproduce the observed groundwater head time series reasonably well. However, we note that there are still some limitations in the current approach, specifically because the current offline-coupling technique simplifies dynamic feedbacks between surface water levels and groundwater heads, and between soil moisture states and groundwater heads. Also the current sensitivity analysis ignores the uncertainty of the land surface model output. Despite these limitations, we argue that the results of the current model show a promise for large-scale groundwater modeling practices, including for data-poor environments and at the global scale.

scholarly journals Integrated mechanistic and data-driven modelling for multivariate analysis of signalling pathways

2006 ◽  
Vol 3 (9) ◽  
pp. 515-526 ◽  
Author(s):  
Fei Hua ◽  
Sampsa Hautaniemi ◽  
Rayka Yokoo ◽  
Douglas A Lauffenburger
Keyword(s):  
Decision Tree ◽  
Mathematical Models ◽  
Large Scale ◽  
Therapeutic Interventions ◽  
Data Driven ◽  
Signalling Pathways ◽  
Apoptotic Pathway ◽  
Tree Analysis ◽  
Modelling Framework ◽  
Novel Approach

Mathematical models of highly interconnected and multivariate signalling networks provide useful tools to understand these complex systems. However, effective approaches to extracting multivariate regulation information from these models are still lacking. In this study, we propose a data-driven modelling framework to analyse large-scale multivariate datasets generated from mathematical models. We used an ordinary differential equation based model for the Fas apoptotic pathway as an example. The first step in our approach was to cluster simulation outputs generated from models with varied protein initial concentrations. Subsequently, decision tree analysis was applied, in which we used protein concentrations to predict the simulation outcomes. Our results suggest that no single subset of proteins can determine the pathway behaviour. Instead, different subsets of proteins with different concentrations ranges can be important. We also used the resulting decision tree to identify the minimal number of perturbations needed to change pathway behaviours. In conclusion, our framework provides a novel approach to understand the multivariate dependencies among molecules in complex networks, and can potentially be used to identify combinatorial targets for therapeutic interventions.

Sign in / Sign up

Export Citation Format

Share Document