Modeling Country Indicators

Project Description

The goal of this project is to visualize and predict relationships between world development indicators in the World Bank. Our objectives are as follows:

1. Visualize relationships between various civil indicators
2. Train a network that predicts various metrics of a country using other metrics as input data
3. Use these networks to predict the future metrics of each country

Usage Instructions

1. Clone this repository by running the following on your terminal

$git clone https://github.com/jimzhang629/DSCI400-Project.git

2. Download the required dependencies by running the following while in the cloned repository

$pip install -r requirements.txt

3. To run LSTM.py and generate predictions for a target indicator, import LSTM.py and call LSTM_predictions. The function requires a target indicator (string) and your data (dataframe) as input.
4. To run VectorAutoRegression.py and generate predictions for a target indicator, import VectorAutoRegression.py and call forecast_VAR_filename. The function requires a filepath (string) to the csv containing the data, as well as a target indicator (string). The function also has an optional paramater, granger_lag (integer), which is 5 by default. It does not require the data to have been wrangled yet, as the module automatically does so. This function outputs the forecasted series for the target indicator.

The Jupyter Notebooks under the demo folder are examples from previous iterations of development on how to run code.

Dependencies

For a list of necessary packages and package versions, see requirements.txt.

To download these dependencies locally, run the following on your terminal

$pip install -r /path/to/requirements.txt

Dataset

Data used for this project can be obtained through this API: https://pypi.org/project/wbgapi/

Below is a snippet of sample code to create and populate a data directory locally:

import wbgapi as wb

df = wb.data.DataFrame('SP.POP.TOTL', 'COL', db=2, time=range(2000, 2020))

The code snippet above creates a local dataframe of values from the SP.POP.TOTL indicator for the country of Colombia (whose ISO code is COL), from the WDI database (signified by db=2), within the range of years from 2000 to 2020.