Penguin Prediction

This repository contains code to do the following:

Train a model to predict penguin species based on penguin dimensions.
Offer the above model's predictions as an API.

The purpose of this repository is to act as a teaching tool, providing both an example of the finished product (on the main branch) and a simple scaffold to start building from (partial-penguin-prediction).

The completed app does not contain code for dealing with malformed requests; this is left as an exercise for the reader.

Installation

pip install -r requirements.txt

Running

python app.py

Deployment

The API is currently deployed on Heroku; the files in this repository are those required for it to run in that environment specifically.

Routes

The root URL is https://penguin-predictions-api.herokuapp.com/.

The following endpoints are available:

/ - the root address, which responds to GET requests with a short description of the API.
/predict - the prediction endpoint, which responds to POST requests with a prediction of penguin species.

When sending POST requests to the API endpoint, include the following four numeric fields in a JSON object:

culmen_length_mm
culmen_depth_mm
flipper_length_mm
body_mass_g

The response will be a JSON object with only one key, prediction, containing the species prediction as a string.

Interface

Penguin prediction is also accessible through a UI at /predict_form.

Data source

The data used in this project is sourced from the Palmer Penguins dataset.