This repository is learning code for designing a solution using MLFlow Recipes. MLFlow Recipes is an open-source project developed by the MLFlow community to provide a set of pre-built, tested, and well-documented machine learning (ML) workflows or "recipes" that can be easily adapted to different ML tasks. These recipes are built on top of the MLFlow framework, which is an open-source platform for the complete machine learning lifecycle, including data preprocessing, model training, evaluation, and deployment. In the project, we are going to design a solution for competition Titanic - Machine Learning from Disaster.
Reference
Download or clone this repository.
- Download the dataset in Titanic - Machine Learning from Disaster
- Extract all the files in
./data/input/
folder - Now, you can run the code using
mlflow recipes
!
You can run this code locally or using databricks.
# Create a Python environment
$ python -m venv .venv
$ source .venv/bin/activate
# Install the requirements
$ pip install -r requirements.txt
# Run using: notebooks/jupyter; or
# Run using: notebooks/databricks; or
# Run using: terminal
$ mlflow recipes run --profile local
# Visualize the experiment performance
# Paths according to `profiles/local.yaml`
$ mlflow ui \
$ --backend-store-uri="sqlite:///metadata/mlflow/mlruns.db" \
$ --default-artifact-root="./metadata/mlflow/mlartifacts"
- License MIT
- Created by leomaurodesenv