MLOps Sum 2024 project repository

Team Members

Team Member	Telegram ID
Alexandra Vabnits	@sashhhaka
Ruslan Izmailov	@Nooth1ng
Bulat Akhmatov	@bulatik1337

Running scripts

All scripts should be run from the repository root, to ensure correct paths initialization in scripts.

Initial whole dataset zameen-updated.csv should be stored in folder "data".

Zameen.com House Price Prediction Dataset

Running documentation

Install requirements:

pip install -r requirements.txt

Set env:

export PROJECT_DIR=$PWD
export PYTHONPATH=$PROJECT_DIR/src
export ZENML_CONFIG_PATH=$PWD/services/zenml
export MLFLOW_TRACKING_URI="http://localhost:5000"

Download dataset from Kaggle:

python src/download_kaggle_dataset

How to start ZenML pipeline:

python pipelines/data_prepare.py

How to start Airflow pipeline:

airflow standalone

MLFlow experiments:

mlflow server
mlflow run . --env-manager=local

Giskard validation:

python src/validate.py

Run tests:

pytest tests

Deploy

Deploy Docker container:

cd api

docker build -t my_ml_service .
docker run -d -p 5123:8080 my_ml_service

docker tag my_ml_service sashhhak0/my_ml_service
docker tag my_ml_service sashhhak0/my_ml_service:v1.0
docker push sashhhak0/my_ml_service:latest

Deploy website:

Start mlflow server:

mlflow ui --port 5000

Deploy Flask API:

python3 api/app.py

Deploy Gradio UI:

python3 src/app.py