Disaster-responses-pipeline
Implemented a machine learning pipeline to categorize emergency messages based on the needs communicated by the sender.
Installation
There is no necessary libraries to run the code, except the ones includes in each file. The code runs with no issues using Python 3.6 or newer versions.
Project Motivation
This project was part of the Data Science Udacity Program and needed to be completed in order to obtain a certificate.
File Description
There are 4 directories that are part of this project. Details below:
Data:
-
categories.csv, messages.csv: Contains the message data to process.
-
process_data.py: Contains the ETL process of the csv files below. Reads the datasets, clean the data and then stores it in a SQLite database. Usage:
python process_data.py messages.csv categories.csv [database_name]Example:
python process_data.py messages.csv categories.csv DisasterResponse.db
app:
- run.py: Flask file that runs web app
- template:
- master.html: Main page of the web app.
- go.html: Classification result page of web app.
models:
-
train_classifier.py: Contains the machine learning pipeline, which uses NLTK, scikit-learn's Pipeline and GridSearchCV to output a final model that uses the message column to predict classifications for 36 categories (multi-output classification). Then, exports the resulting model to a pickle file. Usage:
python train_classifier.py [database_name] [model_export_file]Example:
python train_classifier.py DisasterResponse.db classifier.pkl
notebooks:
- ETL Pipeline Preparation.ipynb: ETL process of the data. The process_data.py file is based on this file.
- ML Pipeline Preparation.ipynb: Contains steps needed for the ML pipeline. The train_classifier.py file is based on this file.
Results
The main findings are found in the ETL Pipeline Preparation.ipynb and ML Pipeline Preparation.ipynb where markdown cells where used to walk through all the steps.