Providentia Forecasting
Providentia Forecasting is a service to do demand forecasting focused on the industry sector.
The general idea is that any person could generate forecasts based on multiple time-series algorithms. These algorithms involved the standard ones for time-series forecasting like ARIMA and also more complex ensemble of machine learning models involving Gradient Boosting and Neural Networks.
This project involved the use of multiple languages, like using Javascript in the front-end (Angular.js) and in the backend (Node.js), RabbitMQ to controlling the queue of requests of forecasts, and Python and R for training multiple models, evaluating these multiple models, choosing or combining model and finally, generating forecasts.
See the demo below: