Disaster Response Pipeline Project

A machine learning pipeline to classify text messages from direct source or social media following a disaster.

Table of Contents

1. Installation
2. Project Motivation
3. File Descriptions
4. Deployment
5. Results
6. Licensing, Authors, and Acknowledgements

Installation

The code have the following dependency

• Python3
• Flask
• numpy
• pandas
• sklearn
• catboost
• matplotlib libraries

Project Motivation

This project was a part of Udacity Nano Degree program to build a model for an API that classifies disaster messages.

File Descriptions

For this purpose, we use Figure Eight. Containing text messages, labeled with different categories. There two csv datasets from Figure Eight located in the data direcctory. disaster_categories.csv contains categories of the messages located in disaster_messages.csv the data directory also contains the pre processing script for the datasets. The app folder contains the Flask web app script. Finally the model directory contains the machine learing pipeline and the saved model classifier.pkl

Deployment

1. To preprocess the data run the following command in your data directory. python process_data.py disaster_messages.csv disaster_categories.csv DisasterResponse
2. To Train and save your trained model run the following in your model direcotry python train_classifier.py ../data/DisasterResponse classifier.pkl <grid_search> The grid_search is a boolean [True|False] parameter which decided to implement grid search of hyperparameters or not
3. Run the following command in the app's directory to run your web app. python run.py
4. Go to http://0.0.0.0:3001/

Results

Front Page of the Webapp Message genere distribution Message categories distribution

Licensing, Authors, and Acknowledgements

Credit is due to Udacity Data Scientist NanoDegree Program!