ABSTRACT
Floods have become more common as a result of the ongoing changes in the environment caused by increased urbanisation and global temperature change, causing havoc on lives and property. As a result, it’s essential to identify the factors that cause floods and flood-prone areas, which could be accomplished by playacting Flood Vulnerability Modelling (FSM) victimisation and hybrid machine learning models to urge correct and semi-permanent results that will be used to implement mitigation measures and flood risk management. To begin, feature choice and multi-collinearity analysis were used to confirm the factors’ prophetical capability and inter-relationships. Following that, IOE was used to analyse the association between the flood moving factors’ categories and flooding, yet because of the influence (weight) of each part on flooding, victimisation quantity, and variable mathematics analysis. The load that was obtained was then used to train machine learning models. The instructed models’ performance was evaluated victimisation the well-known area beneath the curve (AUC) and math indicators. According to the data, the DT-IOE hybrid model had the most effective forecast accuracy of eighty-seven percent, whereas the DT had the rock bottom prediction performance of.0 percent, according to the data. The ultimate susceptibleness maps discovered that around twenty percent of the analysis space is very vulnerable to flooding, which human-induced causes have a giant impact on flooding within the region.
