Titanic and Flask
This repository contains code to build a random forest model and deploy as a web service in Flask application. We use the so well known Titanic dataset.
Launching the app
- Clone repository to your local machine:
git clone git@github.com:leschiffres/titanic_and_flask.git
Using Docker
- Open a terminal.
- Path into the app folder
cd titanic_and_flask - Build the image:
docker build -t titanic_app . - Start the container:
docker run --rm -p 9001:9000 titanic_app - Open a browser and go to the address
http://localhost:9001/
Using existing python
In order for the app to be able to run in the host machine,pandas, numpy, flask, sklearn have to be in the list of installed libraries. Then:
- Open a terminal.
- Path into the app folder
cd titanic_and_flask/app/. - Start the flask application by using the command
python3 app.py. - Open a browser and go to the address
http://localhost:9000/
Files Description
Root
- Dockerfile:
- request.py: Uses a service of the flask application, by sending some data of a passenger in json format, and receives the probability of survival.
Building Model Folder
This folder contains everything regarding the building of the random forest model used in the app.
- dataset : Contains the well known titanic dataset that can be found here: https://www.kaggle.com/c/titanic/data
- titanic_preprocessing_ml.ipynb : Contains all the code for the preprocessing of the data, building of the model and storing the random forest model into a pickle file.
- titanic_predict.ipynb : Reads the test.csv data and produces a prediction of survival. It is not needed for the app but I put it here anyway.
App Folder
It contains all the files building the flask application.
- mapping.py: Maps given values into numbers (based on the mapping that was done during the preprocessing) so that the random forest model can be used accordingly.
- app.py: Launches the flask application. This is the core of the web app.