CRISP DM Data Life Cycle

CRISP-DM (Cross Industry Standard Process for Data Mining) is one of the most widely adopted frameworks for structuring a Data Science project. It breaks the work into six phases that are often revisited multiple times rather than followed once in a straight line.

1. Business Understanding

Defining the project objectives and requirements from a business perspective, and translating them into a data problem.

2. Data Understanding

Collecting initial data and exploring it to get familiar with quality issues, patterns, and interesting subsets.

3. Data Preparation

Cleaning, transforming, and structuring the raw data into a final dataset ready for modeling - this phase usually takes the most time.

4. Modeling

Selecting and applying appropriate modeling techniques, and tuning their parameters to achieve good results.

5. Evaluation

Assessing whether the model actually meets the business objectives defined in the first phase, not just whether it performs well statistically.

6. Deployment

Putting the finished model into a real-world environment where it can be used to support ongoing decisions.

CRISP-DM is deliberately drawn as a cycle rather than a straight line - insights from evaluation often send you back to data preparation or even business understanding, and that back-and-forth is a normal, expected part of the process.

Coming Up Next

Next, you'll look at the different types of data you'll encounter while working through this life cycle.