https://github.com/markdouthwaite/sklearn-flask-api-demo
This repository provides example code for deploying a Scikit-Learn model as a containerised Flask app with Gunicorn. It includes example pytest & locust testing files.
https://github.com/markdouthwaite/sklearn-flask-api-demo
Last synced: 7 months ago
JSON representation
This repository provides example code for deploying a Scikit-Learn model as a containerised Flask app with Gunicorn. It includes example pytest & locust testing files.
- Host: GitHub
- URL: https://github.com/markdouthwaite/sklearn-flask-api-demo
- Owner: markdouthwaite
- License: mit
- Created: 2021-01-21T12:28:20.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2021-02-24T10:19:27.000Z (over 4 years ago)
- Last Synced: 2025-03-19T22:05:08.241Z (7 months ago)
- Language: Python
- Size: 22.5 KB
- Stars: 3
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Scikit-Learn Flask API
This [template repository](https://docs.github.com/en/github/creating-cloning-and-archiving-repositories/creating-a-repository-from-a-template)
provides a minimal configuration for a 'production-ready' Flask API Scikit-Learn
project. It includes a basic project structure and 'seed' files for functional and
non-function testing, a basic application structure (including error-handling
blueprint), and a few assorted 'getting started' files too.
The template has been set up for use with Python >= 3.7 and [Docker](https://www.docker.com/).
## Running locally
To run the basic server, you'll need to install a few requirements. To do this, run:
```bash
pip install -r requirements/common.txt
```
This will install only the dependencies required to run the server.
### Creating a model
Before deploying a model, you'll first need to train a model. There's a demo pipeline
setup in this project. To train the model, run:
```bash
python steps/train.py --path=data/heart-disease.csv
```
This'll produce a couple of files in the `/data` directory, including a pickled
`Pipeline` object. Now you can deploy this pipeline as an API!
### Launching the API
To boot up the default server, you can run:
```bash
bash bin/run.sh
```
This will start a [Gunicorn](https://gunicorn.org/) server that wraps the Flask app
defined in `src/app.py`. Note that [this is one of the recommended ways of deploying a
Flask app 'in production'](https://flask.palletsprojects.com/en/1.1.x/deploying/wsgi-standalone/).
The server shipped with Flask is [intended for development
purposes only](https://flask.palletsprojects.com/en/1.1.x/deploying/#deployment).
You should now be able to send:
```bash
curl localhost:5000/health
```
And receive the response `OK` and status code `200`.
### Querying the model
You can now query the model using:
```bash
curl --location --request POST '127.0.0.1:5000/predict' \
--header 'Content-Type: application/json' \
-d @data/payload.json
```
Where the payload looks like:
```json
{
"sex": 0,
"cp": 1,
"restecg": 0,
"ca": 1,
"slope": 1,
"thal": 2,
"age": 57,
"trestbps": 130,
"chol": 236,
"fbs": 0,
"thalach": 174,
"exang": 0,
"oldpeak": 0.0
}
```
You should see a response looking something like:
```json
{
"diagnosis": "heart-disease"
}
```
And that's it, you have a model wrapped in a web API!
## Running with `docker`
Unsurprisingly, you'll need [Docker](https://www.docker.com/products/docker-desktop)
installed to run this project with Docker. To build a containerised version of the API,
run:
```bash
docker build . -t flask-app
```
To launch the containerised app, run:
```bash
docker run -p 8080:8080 flask-app
```
You should see your server boot up, and should be accessible as before!
With that done, you've got a containerised ML web API ready to go.
## Developing with the template
To develop the template for your own project, you'll need to make sure to [create your
own repository from this template](https://docs.github.com/en/github/creating-cloning-and-archiving-repositories/creating-a-repository-from-a-template)
and then install the project's development dependencies. You can do this with:
```bash
pip install -r requirements/develop.txt
```
This'll install some style formatting and testing tools (including `pytest` and
`locust`).