This example showcases using the Splitgraph API from within Python, as well as using Splitgraph to seamlessly switch between two copies of an SQL database.

It will:

• ingest the Iris dataset from a CSV file
• generate a train and test dataset as distinct Splitgraph images
• use scikit-learn to train a logistic regression classifier on one half of the dataset
• compare the performance of the model on different datasets

Build and start up the engine:

export COMPOSE_PROJECT_NAME=splitgraph_example 
docker-compose down -v
docker-compose build
docker-compose up -d
sgr init

Install this package with Poetry: poetry install

Open the notebook in Jupyter: jupyter notebook iris.ipynb