/Disaster_Response_Pipeline

A disater message classifer buillt on random forest

Primary LanguageJupyter Notebook

Disaster_Response_Pipeline

Udacity data science nanodegree project 2

Table of contents

  1. Libraries used
  2. Project Inspiration
  3. File Descriptions
  4. How to run this project
  5. Data Insights
  6. Licensing, Authors, and Acknowledgements

Libraries used

Python version 3.0. dependecies: json, plotly, numpy, pandas, nltk, flask, plotly, sklearn, sqlalchemy, os, sys, re, pickle, argparse

Project Inspiration

Apply data engineering skills to expand your opportunities and potential as a data scientist. In this project I analyze disaster data from Figure Eight to build a model for an API that classifies disaster messages. The project will include a web app where an emergency worker can input a new message and get classification results in several categories.

File Descriptions

app : template has master.html (main page of web app) and go.html (classification result page of web app), also run.py (Flask file that runs app)

data : data files, oringial data are in csv forms, and other database form data are cleaned data

models : saved model for web app and details of ML in train_classifier.py

asset : web app screenshot

ETL_Pipeline_Prepation.ipynb : extract,transfer and load data. Prepared the data for ML analysis, saved data as database instead of csv.

ML_Pipeline_Prepation.ipynb : serval time consuming ML algorithm process. After considering running time, I use AdaBoost instead of Random forest

all.plk : all the ML models I saved, there are 2 random forest model, 2 AdaBoost models

How to Run

  1. The first part of your data pipeline is the Extract, Transform, and Load process. And run ML pipeline to get trained model saved in plk file
python data/process_data.py  --messages_filename 
data/disaster_messages.csv --categories_filename 
data/disaster_categories.csv --database_filename 
data/DisasterResponse.db

python models/train_classifier.py --database_filename 
data/DisasterResponse.db --model_pickle_filename 
models/classifier.pkl
  1. Once your app is running python run.py , if not running, do this cd app then python run.py. Then open another terminal and type env|grep WORK this will give you the spaceid it will start with view*** and some characters after that Now open your browser window and type https://viewa7a4999b-3001.udacity-student-workspaces.com, replace the whole viewa7a4999b with your space id you got. In this case, spaceid I got is view6914b2f4. Press enter and the app should now run for you

Insights

Result web app here.

web app screenshot

web app screenshot

web app screenshot

Licensing, Authors, Acknowledgements

Data : Figure eight data here