Operationalize a machine learning Application

This project focuses on deploying a machine learning application written in flask. The application serves out predictions (inference) about housing prices through API calls.

Pipeline status

Requirements

To run this project you need to have the following installed

• python3.7
• docker
• hadolint

How to use

run locally

First, setup the virtual environment

make setup

Next, install application dependencies

make install

run the application

python app.py

Once the application is running, to make predictions, run the following in a different terminal.

./make_predicitons.sh

run with docker

Run application in a doker container

./run_docker.sh

Make predicitons. ie run it in a different terminal

./make_predictions.sh

run with kubernetes

Push image to docker repo

./upload_docker.sh <docker-username> <docker-password>

Run container in a kubernetes pod

./run_kubernetes.sh

Make prediction

./make_predicitons.sh

Files in this repo

• makefile: contains commands to simplify some of the steps in the project like creating the virtual environmnet, linting etc
• run_docker.sh: this script buids and runs a docker cobtainer using the built image
• upload.sh: script that pushes the afore created image to a specified docker hub repository
• run_kubernetes.sh: creates a secret with the docker information and created a pod using the pushed image
• app.py: python application reponsible for taking in data and making predictions
• requirements.rxt: dependencies of the machine learning app