This is an all level friendly repo showing how machine learning pipeline can be built from scratch adopting the procedural programming approach or custom pipeline code or third-party code leveraging on the sckit-learn library. All three pipelines are built with the Titanic data set from Kaggle in mind https://www.kaggle.com/c/titanic/data.
It is meant to show you how codes from the research environment 'Jupyter Notebook' are gradually been transformed into reusable pipelines while ensuring reproducibility and modularity in mind. I have also organised the code in a way that is easy for you to edit if you want to make changes to any of the file.
pip install pandas==1.18.1
pip install numpy==0.25.3
pip install Scikit-Learn==0.22.1@OlugbamiEzekiel – ezekiel.olugbami@gmail.com
https://github.com/ezekielolugbami
Fork it (https://github.com/ezekielolugbami/ml_pipeline_from_scratch.git)