/Disaster-Management-ML

Primary LanguageJupyter Notebook

Disaster Response Pipeline Project

Overview

This project is in a form of a Web App implemented using python and its libraries, ETL techniques, Machine Learning Models and Flask App to visualize a system that can classify messages sent during disasters and correlate it to various categories in order to facilitate disaster response. This is a Udacity Nanodegree Educational Project.

Contents

The repo consists folders of the following structure:

  • app

    • template
      • master.html       # main page of web app
      • go.html       # classification result page of web app
    • run.py       # Flask file that runs app

  • data

    • disaster_categories.csv       # data to process
    • disaster_messages.csv       # data to process
    • process_data.py       # python script for the ETL Transformation of the data
    • DisasterResponse.db       # database created from code to save cleaned data

  • models

    • train_classifier.py       # python script to create machine learning pipeline model for the dataset
    • classifier.pkl       # saved model

  • notebooks

    • ETL_Pipeline_Preparation.ipynb       # Jupyter Notebook to explain ETL Script step by step
    • ML_Pipeline_Preparation.ipynb       # Jupyter Notebook to explain ML Model creation step by step

  • README.md

Code Flow and Results

The code cleans data extracted from dataset containing real messages that were sent during disaster events using process_data.py file. Then, it runs a ML Pipeline using CountVectorizer, TfidfTransformer then MultiOutputClassifier and KNeighborsClassifier to train the model.

After that, GridSearch is used to find the best parameters for the classifier and the results was using 10 n_neighbours and uniform weights. To enhance the model further, a custom extractor called StartingVerbExtractor was used to increase the accuracy.

The precision reached 0.7 after being 0.58 before enhancement.

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

  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/