Since a crisis is linked to a country and all countries are connected to each other, we have to build a graph to see how much the crisis in one country will affect the other.
What it does
Implementation of a web application that will allow to predict a banking crisis of a country at a given year with its different partners.
How we built it
- Use the WITS API to import our data,
- Preprocess the data on COLAB using the preprocessing libraries (pandas and scikit-learn),
- Visualize the data to extract information to predict a crisis using the visualization libraries (seaborn and matplotlib),
- Import the data sets obtained in TigerGraph and then apply the k-nearest neighbors classification model that TigerGraph makes available to us,
- Predict the countries in crisis according to the year and list for a country these partners affect and
- Set up a visualization interface with Plotly Dash.
Challenges we ran into
This was the first time we used tigerGraph ,so it wasn't easy to manage, but we did it anyway.
Achievements we are proud of
We are able to do these things with tigerGraph:
- Identify duplicates, missing values in a static news dataset in the graph,
- Predict if a country will be in crisis at a specific year and based on the data entered,
- Identify the countries that will be impacted by the crisis of a country, and
- Perform and visualize all this with a graphical interface.
What we learned
We learned that the graph database works faster than the others, and truth be told, we didn't know we could use machine learning algorithms in tigerGraph until we found out that it included everything about graph analysis, which is awesome.
What's next for Predict Global Crises
In the future, we intend to set up a system to predict the resources (material, financial and food) to be allocated to countries in crisis.