What is an ML pipeline?
Machine learning (ML) pipeline is a way to automate the workflow of machine learning model creation. ML pipeline consists of all the steps it takes to create and deploy a machine learning model. The steps are: data gathering/extraction, data preparation/preprocessing, feature engineering, model training, model evaluation, model deployment.
Why is it important?
ML pipeline allows data scientists to separate each stage of the workflow into an independent service. They can be used all together or separately, depending on the needs. All the steps are standardized, which means that no additional changes have to be made to other services in case of any changes in one of them.