ML Life Cycle

You will learn: the end-to-end process to develop an ML product.

We have finally arrived to the point of grouping all the foregoing pieces to be able to successfully understand how the Machine Learning Life Cycle can be managed and applied towards the creation of a new ML product.

In the following lectures, we will review each of the life cycle stages:

Business ObjectivesData PreparationModel DevelopmentTrain, Evaluate, and DeployA/B TestingKPI Evaluation

Some considerations

Before we start defining each stage, let's first take into account that depending on the product stage we are at, the ML life cycle will require certain adjustments.

As a reminder, we can split our product initial stages as:

• 🚀 MVP (Minimum Viable Product): the first launched version of your product to customers. It is developed with a minimum number of features to solve your problem in a simple way. It is generally used as a trial tool only.

• ❤️ MLP (Minimum Lovable Product): similar to an MVP, but with special consideration for its design and UI. It aims to solve the problem, but also delight users.

• 💰 MMP (Minimum Marketable Product): this is the version of your MVP (or MLP) which you will launch to market.

Readings

https://blogs.oracle.com/datascience/7-artificial-intelligence-trends-and-how-they-work-with-operational-machine-learning-v2