Context
In the aftermath of disasters, one of the challenges in societies is to decide rapidly on the type of response and emergency support to provide to victims.
Data science and machine learning can help speed up the decision process and help responders send adequate emergency responses.
Dataset
Appen generated a dataset containing over 26,000 emergency messages classified into 36 response categories.
Approach
I leveraged Natural Language Processing and built a Random Forest multi-class classifier to classify any emergency message from disasters.
This classifier can be viewed in a local Flask App, along with a visual summarization of the Appen training dataset.
Tools
Python, Flask
Isabelle Vea PhD 