Get your score in minutes
Answer a series of yes or no questions intended to examine where risks and opportunities are within the four stages of the data science lifecycle (Manage, Develop, Deploy, Monitor). After completion, you’ll immediately receive a Model Velocity Assessment with areas of improvement and actionable recommendations.
Frequently Asked Questions
Will I be able to share my results?
Yes. Once you have completed the free assessment, the results page with your Model Velocity Score will include a unique URL that you can share with others.
What is model velocity?
Model velocity is the aggregate rate at which a company is able to perform all of the steps in the data science lifecycle. Companies become more model-driven as they increase their model velocity. This conceptual goal is focused on optimizing technology, people, and process.
In addition, model velocity includes the velocity to get access to the tools and compute you need, the velocity to onboard new data scientists, the velocity to re-use prior work, the velocity to experiment, the velocity to publish, the velocity to identify degraded models, the velocity to retrain, and so on.
What are the phases of the data science lifecycle?
The data science lifecycle has four phases: Manage, Develop, Deploy, and Monitor. All must operate efficiently and at scale in order to achieve high model velocity.
- Manage: A business problem needs to be improved with insights from data or models. These problems are prioritized and scoped so the data science team can begin work. Models needing refresh or retraining are part of the prioritization process. The team reviews prior work and potential data sources that can be leveraged in the project
- Develop: Model/ development includes identifying and accessing data, preparing it for use and the creation of models/analysis to solve the business problem. Data scientists collaborate to create the best model to solve the problem. It may take 100’s of iterations using different tools to find the best solution.
- Deploy: Validation and testing are necessary prior to deployment/use to ensure the model/analysis performs as expected. Then it is placed into a system or process for use.
- Monitor: Continual monitoring of models ensures they perform within expected parameters. If performance decays, they should be refreshed, retrained or replaced quickly.
How long is the data science process?
The goal for the data science lifecycle is to shorten it as much as possible while still ensuring high-quality, governed models. Data science is research at its heart—it’s experimental and iterative so you may try dozens or hundreds of ideas before getting something that works. Moving efficiently through those iterations and the validation and deployment processes is critical because the longer it takes to complete and deploy a model, the higher the opportunity cost and the likelihood that the model has begun to decay.
Who is involved in the data science process?
The data science lifecycle encompasses four phases that provide a thumbnail sketch of the overall process and indicate where different team members should be focusing (manage, develop, deploy, monitor). However, the roles and responsibilities in the typical lifecycle are seldom this clearly defined. Read more about the 7 key roles in MLOps, or machine learning operations, typical of enterprise organizations delivering models. In short, the 7 roles are: data scientist, data analyst, data engineer, devops engineer, ML architect, software developer, and domain expert/business translator.
Is one step in the data science process more important than the others?
It depends on your perspective. From a high quality model perspective, each step is critical and equally important. From a business perspective, the deployment step is the critical point where tangible business value is created.
What is Domino Data Lab?
Domino’s enterprise MLOps platform accelerates research, speeds model deployment, and increases collaboration for code-first data science teams at scale. Over 20% of the Fortune 100 has chosen Domino to unleash data science using powerful and unbeatable capabilities in data science development, collaboration, project management, and model publishing.