Assessment of Business Interruption of Flood-Affected Companies Using Random Forests

Abstract

Losses due to floods have dramatically increased over the past decades, and losses of companies, comprising direct and indirect losses, have a large share of the total economic losses. Thus, there is an urgent need to gain more quantitative knowledge about flood losses, particularly losses caused by business interruption, in order to mitigate the economic loss of companies. However, business interruption caused by floods is rarely assessed because of a lack of sufficiently detailed data. A survey was undertaken to explore processes influencing business interruption, which collected information on 557 companies affected by the severe flood in June 2013 in Germany. Based on this data set, the study aims to assess the business interruption of directly affected companies by means of a Random Forests model. Variables that influence the duration and costs of business interruption were identified by the variable importance measures of Random Forests. Additionally, Random Forest-based models were developed and tested for their capacity to estimate business interruption duration and associated costs. The water level was found to be the most important variable influencing the duration of business interruption. Other important variables, relating to the estimation of business interruption duration, are the warning time, perceived danger of flood recurrence and inundation duration. In contrast, the amount of business interruption costs is strongly influenced by the size of the company, as assessed by the number of employees, emergency measures undertaken by the company and the fraction of customers within a 50 km radius. These results provide useful information and methods for companies to mitigate their losses from business interruption. However, the heterogeneity of companies is relatively high, and sector-specific analyses were not possible due to the small sample size. Therefore, further sector-specific analyses on the basis of more flood loss data of companies are recommended.

The object-specific flood damage database HOWAS21

“…categorical and continuous data, they allow for non-linear and non-monotonous input data, and they are able to capture predictor interactions (e.g. Merz et al, 2013;Schröter et al, 2014;Kreibich et al, 2016;Sieg et al, 2017;Sultana et al, 2018).…”

Random Forest Ability in Regionalizing Hourly Hydrological Model Parameters

“…Although the RF algorithm has demonstrated its outperformance in comparison with other ML algorithms, and despite its application to many problems in several environmental sciences [34][35][36], its use in water sciences is still limited [30]. Examples of RF applications in hydrology include precipitation downscaling [37,38], flood prediction and risk assessment [39][40][41], estimating runoff modes or hydrological signatures on a continental scale and predicting flow regimes [42][43][44][45][46] as well as predicting flow characteristics at ungauged locations [47][48][49]. RFs are constructed by growing a number of regression and classification trees.…”

Subdaily Rainfall Estimation through Daily Rainfall Downscaling Using Random Forests in Spain

“…Random forests (RF) [54] were used as the machine learning regressor. RF has been used and evaluated in several water resources studies [55][56][57]. An advantage of RF is that it does not require any log or logit transformation of the input to constrain the predicted results, simplifying the application of the method.…”