tenniskit / project-2-disaster-response

To find the related categories for a custom message sent by users.

Disaster Response Pipeline Project

Table of Contents

1. Introduction
2. What's included
3. File Descriptions
4. Installation
5. Instructions
6. Usage
7. Licensing, Authors, and Acknowledgements

Introduction

The project aims at providing a web application called "Disasters" for end-users to enter text message during disaster, and it is able to classify the text message into 36 categories.

What's included

(project folder)/
├── data/
│   ├── disaster_categories.csv
│   ├── disaster_messages.csv
│   ├── process_data.py
│   ├── output.png
│   ├── (DisasterResponse.db)
├── models/
│   ├── train_classifier.py
│   ├── output.png
│   ├── (classifier.pkl) 
└── app/
    ├── run.py
    ├── output.png
    ├── templates/
        ├── go.html
        ├── master.html

File Descriptions

1. data/disaster_categories.csv
   • Categories dataset from Figure Eight.
2. data/disaster_messages.csv
   • Messages dataset from Figure Eight.
3. data/process_data.py
   • Python script file for ETL pipeline.
4. data/output.png
   • Screenshot of process_data.py output.
5. data/DisasterResponse.db
   • The output database file combining Categories and Messages datasets. This file is not initially included in the project, but generated after executing process_data.py.
6. models/train_classifier.py
   • Python script file for ML pipeline.
7. models/output.png
   • Screenshot of train_classifier.py output.
8. models/classifier.pkl
   • The saved model generated by train_classifier.py.
9. app/run.py
   • The web application "Disasters" for end-user to enter custom message and get classification result.
10. app/output.png
    • Screenshot of web application "Disasters".
11. app/templates/go.html
    • HTML template file used by the web application "Disasters".
12. app/templates/master.html
    • HTML template file used by the web application "Disasters".

Installation

The code should run with no issues using Python versions 3.6.3

Python libraries used in the project:

1. pandas
2. numpy
3. sqlalchemy
4. re
5. nltk
6. nltk.tokenize
7. nltk.stem
8. sklearn.model_selection
9. sklearn.pipeline
10. sklearn.base
11. sklearn.feature_extraction.text
12. sklearn.ensemble
13. sklearn.multioutput
14. sklearn.metrics
15. json
16. plotly
17. plotly.graph_objs
18. flask
19. joblib
20. sys

Instructions

1. Run the following commands in the project's root directory to set up your database and model.

   • To run ETL pipeline that cleans data and stores in database python data/process_data.py data/disaster_messages.csv data/disaster_categories.csv data/DisasterResponse.db

     

   • To run ML pipeline that trains classifier and saves python models/train_classifier.py data/DisasterResponse.db models/classifier.pkl

     Please note that ML pipeline takes time to train model. In a testing machine with Intel i7 CPU and 32GB ram, it took 1.5 hours to complete.

2. Run the following command in the app's directory to run your web app. python run.py

3. Go to http://localhost:3001/

   

Usage

1. Enter the custom message in the textbox. Then hit "Classify Message".

   

2. It will show all categories that the custom message belongs to with green highlighted.

   

Licensing, Authors, Acknowledgements

Code released under the MIT License. Must give credit to Figure Eight for the data.