Disaster Response Web Application is a Web app that can help emergency organizations analyze incoming messages and classify the messages into specific categories (e.g. Water, Food, Hospitals, Aid-Related) during a disaster event. The app is based on Nature Language Processing and Random Forest Classifier ML model. The data was collected by Figure Eight and provided by Udacity.

The techniques I used in this project include:

• SQLite in Python
• Bag of Words (BOW)
• Multilabel classification
• Build Machine Learning Pipeline
• Grid Search Cross Validation
• Random Forest Classifier
• K nearest neighbors (KNN)
• Build a Flask web app

You can find the full analysis in my GitHub repo.

Data

The data files associated with this project are from Figure Eight

• messages.csv: FIgure Eight provide 26,248 messages
• categories.csv: Raw categories data, total 36 categories.

Project Process

1. ETL Pipeline

   • Loads messages and categories dataset
   • Clean data
   • Feature Engineering
   • Stores it in a SQLite database

2. ML Pipeline

   • Loads data from the SQLite database
   • text processing and machine learning pipeline
   • Trains and tunes a model using GridSearchCV
   • Exports the model

3. Build web app

   • Create HTML templates
   • Build Flask web application

Folder Structure

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20

├── README.md          
│
├── models                   <- Trained models and ML pipeline
│   ├── classifier.pkl       <- Saved model
│   └── train_classifier.py  <- Scripts to train model sdf
│
├── requirements.txt         <- File for reproducing the environment
│
├── data                     <- Raw and processed Data; data     
│   │                           cleaning script
│   ├── messages.csv         <- Raw messages data
│   ├── categories.csv       <- Raw categories data
│   ├── DisasterResponse.db  <- Saved processed data
│   └── process_data.py      <- Scripts to process data
│
├── notebooks                <- Jupyter notebooks
│
└── App                      <- Source code for use in this project.
    ├── templates            <- Flask html templates 
    └── run.py               <- Scripts to create start Flask server. 

Full Instructions

• Installation Install Python 3.5+ Run pip install -r requirements.txt

• Prepare data

  1. Clone or download the repo
  2. Open terminal and navigate to the project folder
  3. Run python data/process_data.py data/disaster_messages.csv data/disaster_categories.csv data/DisasterResponse.db

• Train model 5. Run python models/train_classifier.py data/DisasterResponse.db models/classifier.pkl

• Start Web app 6. Run cd app 7. Run python run.py 8. Open web browser and go to http://loclhost:3001 or http://0.0.0.0:3001

Screen Shot

If you think your friends/network would find this useful, please share it with them – I’d really appreciate it.

Facebook

Twitter

E-Mail

LinkedIn

Reddit

Telegram