goodbsw/disaster-response-pipeline

Disaster Response Pipeline

Introduction

This project is for categorizing 36 disaster classes(e.g. flood, strom, shelter, ...) based on the messages delivered during disasters. The training set of data is provided by Figure Eight to build a model for an API that classifies disaster messages. The project has been developed as part of Udacity Nano degree for data scientist.

The project contains:

1. ETL pipeline to wrangle data and load in SQLlight database.
2. Machine Learning pipeline to train classifier and optimize the classifier hyperparameters using GridSearchCV
3. Web application to visualize train data and show result of trained classifier

File description

data:

• data/proces_data.py: python script to clean data to save in SQLlite database
• data/disaster_categories.csv and data/disaster_messages.csv: csv files with train data

model:

• model/train_classifier.py: python script to train classifier using machine learning algorithms and sklearn pipeline

app:

• app/run.py: python script contains api with data visualization using plotly
• templates/: directory that contains html for each api

Usage

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

• Read the data from csv files to run ETL pipeline to create DisasterResponse.db SQLlite database

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

• Read the data from the database to train classifier

python app/run.py

• Run web application

Webapp description

1. Three types of messages' genres

2. The distribution of message categories

3. The result of message search