In this project we are using a time series model to understand the patterns of natural disasters declarations in the United States. By understanding these patterns we will then provide recommendations on what natural disasters to prepare for and when they may occur.
The dataset used is an OpenFEMA Dataset for Disaster Declarations and can be found at the following source: https://www.fema.gov/openfema-data-page/disaster-declarations-summaries-v2
Using ARIMA Modeling I predicted the future of top three natural disasters to most commonly occur in the U.S. This modeling looked at the occurance of natural disaster up to 36 months in the future from 01-01-2020. The ARIMA modeling allowed us to create the following prediction of the occurance of the top three natural disasters.
The top three natural disasters in the U.S.A since 1970 that I studied are:
- Severe Storms
- FEMA should prepare communities for severe storms at highest risk before May in following years
- Hurricanes
- FEMA should prepare communities for severe storms at highest risk in August to October
- Floods
- FEMA should prepare communities from March to June for increased flooding occurrence and damagesSevere Storms - Slow, but consistent occurrence of Severe Storms in the Future
With greater time and scope I would pursue the following additions to the project:
- Research areas of the U.S.
- Biological Natural Disasters
- COVID-19 & other possible pandemics
- Snow, Fire, & Ice
- Funding allocation