Udacity data science nanodegree project for classifying disaster messages with machine learning

Motivation:

This project has been conducted as a part of Udacity's Data Scientist Nanodegree. The Students were to build:

• An ETL pipeline to load csv files, transform and clean data and load to a sql database.
• A Machine learning pipeline
• A web app that puts it all togheter and displays some visualisations for the data and predict categories for new messages

Web

The complete project can be seen on www.brønstad.com/disaster

Installations:

• python 3.6
• Flask
• numpy
• pandas
• sklearn
• nltk
• sqlalchemy
• re
• seaborn
• pickle

Instructions

Run the following commands in the root directory.

1. Run the ETL pipeline to load data from csv files, clean and load into a database as specified in the third arg

• python data/process_data.py data/disaster_messages.csv data/disaster_categories.csv data/DisasterResponse.db

2. Run a ML pipeline to trains and saves a model

• python models/train_classifier.py data/DisasterResponse.db models/classifier.pkl

• Run run.py and go to http://0.0.0.0:3001/

Files:

The project contains a jupyter notebook for the ETL pipeline aswell as on for the ML pipeline. a folder with data, 1 folder for the webapp and a folder for the model

Licensing, Authors, Acknowledgements, etc.

Thanks to Figure Eight for the dataset