BPS: Population Forecasting

Using a novel nested deep-learning neural network with multivariate polynomial regression models, we can predict the population of Southeast Asian countries based on relational data.

Comment

Inspiration

In the future, Southeast Asian countries will continue to develop at alarming rates and be forced to adapt their government policies and actions to accommodate for the increasing populations in terms of infrastructure, resources, and economic stability.

What it does

Our model is built in two phases: one for population prediction based on multiple variables including birth rate, infant mortality rate, fertility rate, net migration, real GDP per capita, and the youth proportion of the population. Using an artificial neural network, we trained the model to predict population based on these 7 variables. However, we eventually realized that we had to develop another regression model because there is no way for us to know the values of these variables in the future. Instead, we have multiple regression models for each of the features such that using a country and year, every feature is predicted and, using the predicted features, a population growth is predicted.

How we built i

Built With

Share this project:

Updates