Document Type



Master of Science (MS)



First Advisor's Name

Wensong Wu

First Advisor's Committee Title

committee chair

Second Advisor's Name

B. M. Golam Kibria

Second Advisor's Committee Title

committee member

Third Advisor's Name

Florence George

Third Advisor's Committee Title

committee member


predictive model, k-fold cross-validation, variable selection, prediction performance

Date of Defense



The impacts of disaster on the private sector are inevitable, but their risks can be managed and reduced by preventively evaluative measures. Disaster risk reduction index (DRRI) and Disaster Experience (DE) variables were investigated in a survey study in six Western Hemisphere cities within the private sector of various business sizes. Our thesis built and evaluated 16 predictive models of DRRI with 36 categorical predictors and N = 1162 observations. Four statistical methods for linear regression and five for classification as well as seven machine learning methods were utilized. We also used stepwise selection and regulation methods for variable selection. They improved the performance of some models. To evaluate and compare the prediction performance among all models, we used resampling 5-fold cross-validation (CV) to estimate the true mean squared error (MSE) and classification accuracy. The results indicated that the neural network was outperformed among all the predictive models with the highest classification accuracy.





Rights Statement

Rights Statement

In Copyright. URI:
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).