Machine Learning Pipelines

What is it? A machine learning pipeline is a set of steps that are used to build and deploy a machine learning model which usually includes 8 steps:

1. Problem Definition: Define the business problem
2. Data Ingestion: Identify and collect the dataset
3. Data Preparation: Process and prepare the data
4. Data Segregation: Split data into training/validation/testing set
5. Model Training: Train the models against the training dataset
6. Candidate Model Evaluation: Measure the performance of the models
7. Model Deployment: Deploy into production
8. Performance Monitoring: Monitor performance, retrain and calibrate