*6.2. Future Work*

Further extensive studies are needed, and recommendations for future research are discussed as follows. First, this study analyzes the importance of features that affect seismic mortality, which simply collects 14 features and classifies them into disaster-inducing factors, disaster-affected bodies and disaster-formative environments. Future studies can extend the research by refining the classification standard and increasing the number of factors. Second, this study divides the study area into risk zones of three grades based on regional differences, where the partition standard exerts a potential influence on the accuracy and applicability of the proposed model. Future studies can explore more reasonable criteria for different study areas. Third, the proposed prediction approach is a regression model that is based on SVR, which is essentially a data-driven model. Future studies can build models based on deeper seismic mechanisms to predict deaths that are caused by earthquakes.

#### **7. Conclusions**

This study evaluated the importance of 14 features that affect seismic fatality based on the RF model. On the basis of the importance assessment, we selected magnitude, population density, geological fault density and GDP as the input parameters of the prediction model, among which the densities of population and geological faults were also integrated for spatial division. This study also proposed a spatial division method based on the theory of regional difference. We studied the regional diversity of geological fault density and population in China's mainland using the WorldPop population dataset (100 m resolution) every five years from 2000 to 2020 and the strata fault line dataset and, finally, divided the study area into zones of various risk grades by overlay analysis. Based on the results of feature selection and spatial division, this study proposed a zoning prediction model based on SVR. Using 113 samples in the earthquake case dataset, we implemented model training and obtained the optimal model parameters for each risk zone to enhance the prediction accuracy of earthquake death tolls. The following conclusions were drawn from the results that were obtained in this study:


**Author Contributions:** B.L. and T.Z. implemented the research and wrote the original manuscript. A.G. provided the original idea for the study and supervised the research. W.B., C.X. and Z.H. aided with the manuscript revision. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was jointly supported by the National Key Research and Development Program of China (Grant No. 2019YFE01277002, No. 2017YFB0504102 and No. 2017YFC1502704) and the National Natural Science Foundation of China (41671412).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Our research data are from relevant open data websites, which can be obtained according to the links listed in our references.

**Acknowledgments:** The authors would like to express deep gratitude to Jianghong Zhao from Beijing University of Civil Engineering and Architecture for her guidance on the framework design of the paper.

**Conflicts of Interest:** The authors declare no conflict of interest.
