ML-Pipeline (EN)

Definition

The ML-Pipeline is a structured process that encompasses all steps from data collection to model deployment. It enables systematic and reproducible development of ML applications. Each phase of the pipeline has specific requirements and objectives that are crucial for the success of the overall project. A well-designed pipeline ensures that data, models, and results can be efficiently managed.

Architecture

flowchart TD     A[Data Collection] --> B[Data Cleaning]     B --> C[Feature Engineering]     C --> D[Model Training]     D --> E[Model Validation]     E --> F[Model Deployment]     F --> G[Model Monitoring]     G --> H[Feedback Loop]     H --> C 

In Context

• Typically used together with MLOps platforms
• Related to: Data Pipeline, Feature Store, Model Registry
• Example use: Production environments for recommendation systems or prediction models