The solution to Problem E of MCM in 2018.
An evaluation model of State Vulnerability using nerual networks(Pytorch implementation).
I have selected 12 indicators such as GDP, CPI, temperature and precipitation to evaluate the vulnerability of a country.
The training data set comes from World Bank statistics.
Therefore,by collecting a country's performance on these 12 indicators, we can judge the country's vulnerability.
I use the data of 168 countries in 2016 to train neural networks.
After my training, the model has performed very well and is able to fit the data in the data set.
I used the data of 2017 to test the network and the network showed a good generalization ability.The training and testing data is saved in the csv format in the ./scripts/data folder.
1、Predicting national vulnerability through the value of indicators
2、Analyze the impact of certain indicators on the vulnerability of the country
3、Learning data reading, model building, and fitting other functions
4、And anything you like...
Please confirm that you have installed the following third party packages：
1、Pytorch0.2(or later)
2、numpy1.14.1
3、pandas0.22.0
3、matplotlib2.2.2
You may install these by:
>> pip install requirements.txt
After clone this repo,then:
>> cd country_vulnerability
Then:
>> python scripts/main.py
If you have any questions or good ideas about this project, you can send me an email.
Apache License 2.0