DisasterResponsePipeline

Project of Udacity Data Scientist Nanodegree Program

Table of Contents

• DisasterResponsePipeline
  • Table of Contents
  • Installation
  • Project Motivation
  • File Descriptions
  • Results and Discussion
  • Licensing, Authors, Acknowledgements

Installation

Required packeages are listed in requirement.txt.

Follow follows steps to run the app:

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

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

3. Go to http://0.0.0.0:3001/

Project Motivation

In this project, the process of a comprehensive implementation of Machine Learning in realworld project is demonstrated, which includes following steps:

1. ETL process includes extracting data, cleanning data and storing the clean data into a SQLite database.
2. Using NLP, Pipeline and GridSearchCV to classificate data.
3. Deployment the model as a web app

File Descriptions

There are 3 directories here.

1. Directory app contains the script to start the web app run.py and the webpage templates in subdirectory templates
2. Directory data contains the origin data disaster_categories.csv and disaster_messages.csv, the ETL script process_data.py and the database DisasterResponse.db which saves the cleaned data.
3. Directory models saves the ML script train_classifier.py and the saved trained ML model final_model.py.

Results and Discussion

It should be pointed out, there is still much room for imporvement. An obvious problem is the data is imbalanced, which has stongly influenced the accuracy and precison of the trained model. Another improvement is to employ the model in a cloud server rather than locally. Finally, due to the restriction of computation power, GridSearchCV here is only to demonstrate the pipeline to employ it rather than to provide optimized trained results.

Licensing, Authors, Acknowledgements

Must give credit to Udacity for the project. Otherwise, feel free to use the code here as you would like!