DataTerminology.comNEW
Also known as: ML lifecycle, Project lifecycle
Framework of stages guiding data projects from problem framing and data preparation through modeling, deployment, and continuous monitoring.
DESCRIPTION
The data science life cycle structures work into phases such as business understanding, data acquisition, preparation, modeling, evaluation, deployment, and monitoring. CRISP-DM and TDSP are common frameworks guiding best practices. Lifecycle discipline aligns teams, clarifies deliverables, and mitigates risks. MLOps extends lifecycle management to production systems with CI/CD, model registries, and drift monitoring. Effective documentation and governance ensure repeatability and compliance.
ALTERNATIVE DEFINITIONS
Microsoft Learn (TDSP)
The Team Data Science Process (TDSP) provides a lifecycle to structure data science projects.
Presents a lifecycle model with stages and roles tailored to team workflows.
IBM
CRISP-DM is a data mining process model that describes the life cycle of data mining projects in six phases.
Classic six-phase process widely adopted across industry.
Google Cloud
MLOps is a set of practices for collaboration and communication between data scientists and operations professionals to help manage production ML lifecycle.
Focuses on production lifecycle and operational practices for ML systems.
RELATED TERMS
MLOps Experiment Tracking Model Deployment Feature Engineering Data Capture A/B Testing Model Monitoring Model Registry Governance Deployment Pipeline
COMPETING TERMS
KNOWLEDGE AREAS
Data Science MLOps AI Testing & Validation Collaboration Data Management
TARGET AUDIENCE
Data Scientist MLOps Engineer AI Product Manager Data Engineer Data Strategy Consultant Chief Data Officer
EXTERNAL SOURCES
- Wikipedia - Cross-industry standard process for data mining
- Microsoft Learn - TDSP lifecycle overview
- IBM - CRISP-DM process guide
- Google Cloud - Production ML lifecycle
- MLflow - Experiment tracking tools
TOP QUESTIONS
Why is a Data Science Life Cycle important?
It provides structure, reduces risk, and aligns stakeholders by defining clear phases, deliverables, and checkpoints.
Category: Relevancy
How does TDSP differ from CRISP-DM?
TDSP emphasizes team roles and agile iteration; CRISP-DM formalizes six phases for data mining projects.
Category: Opinion
What are common mistakes in lifecycle management?
Skipping stakeholder alignment, underdocumenting assumptions, and neglecting monitoring lead to project failures.
Category: Mistakes
How does MLOps extend the lifecycle?
It operationalizes deployment, monitoring, CI/CD, governance, and drift management for production ML systems.
Category: Overview
What is required to implement a lifecycle effectively?
Clear roles, templates, tooling for tracking, and governance policies ensure repeatable workflows.
Category: Requirements
What’s the biggest risk when ignoring lifecycle stages?
Models may be unreliable, unmaintained, and misaligned with business needs, causing wasted effort.
Category: Risks
How to evaluate models within the lifecycle?
Use holdout validation, cross-validation, A/B testing, and monitoring of performance and drift post-deployment.
Category: Testing
Where should teams start?
Adopt CRISP-DM or TDSP templates, define metrics, and integrate experiment tracking tools like MLflow.
Category: Learning
What is a hot topic around the lifecycle?
Integrating responsible AI practices—bias checks, explainability, and model cards—into each phase.
Category: Topic