Detect Data Drift

Motivation

Data drift occurs when the distribution of input features in the production environment differs from the training data, leading to potential inaccuracies and decreased model performance.

To mitigate the impact of data drift on model performance, this workflow automates the process of detecting drift, notifying the data team, and triggering model retraining.

Try it out

Clone the repo:

git clone https://github.com/khuyentran1401/detect-data-drift-pipeline

Next, create and start a Docker container running a PostgreSQL server with prepopulated tables:

docker run -d -p 5432:5432 -e POSTGRES_PASSWORD=... -e POSTGRES_USER=... khuyentran1401/bikeride-postgres:latest

Before running the application, add the required environment variables to the ".env" file:

POSTGRES_USERNAME=...
POSTGRES_PASSWORD=...
SLACK_WEBHOOK=...
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY_ID=...

Encode these environment variables and save them in the ".env_encoded" file by running:

bash encode_env.sh

Now, start the containers for Kestra:

docker compose up -d

You can access Kestra's user interface at http://localhost:8080.

To import example flows into Kestra, click the "Import" button and select the files located in the "kestra_pipeline" directory

After importing, you will see the following flows: