The Application of Using Supervised Classification Techniques in Selecting the Most Optimized Temporary House Type in Post-disaster Situations

Abstract

The United States spends around 450 million dollars just for lodging the survivors and creating temporary shelters in the wake of hurricane Harvey, Irma, and Maria in 2017. Post-disaster temporary housing is a multi-objective process, thus reaching the optimized model relies on numerous objectives and their interaction with each other. The construction activities, especially in a post-disaster scenario, is considered challenging, leading to ineffective management in post-disaster housing reconstruction. Acknowledging and creating a balance between these issues by the policymakers who provide accommodations to post-disaster victims is one of the main challenges that need to be addressed.Post-disaster temporary housing is an integral part of the recovery process; however, not many research types have been done regarding how the factors and to what degree factors can affect the classification. One way to categorize the temporary housing units (THU) is based on how they get built, either made onsite or created offsite. However, the mechanism of selecting the THU types is mainly based on expert opinion and tacit knowledge, which can result in the insufficiency of the process.This model aims to study how and to what degree the factors that affect the post-disaster temporary housing process dictate the type of THU in terms of being built onsite or modeled offsite. The researchers designed a questionnaire to understand each main factor's importance compared to each other through a ranking process. It also asks each participant to rate the different THUs being used based on their importance. In this study, an ordinal classification framework is introduced using the K-Nearest Neighbor (KNN) model to help decision-makers choose the right type of temporary houses based on their needs. This model's results show how supervised classification models can be an efficient tool and holistic approach to providing more robust, efficient decisions as an alternative to the current strategy, which relies on tacit knowledge.