Segmentation is used on every project

Every structural analysis will include segments of the overall data. The overall data by itself is not interesting from an analysis perspective, it is only when you break down each segment that you can learn what drives the business.

Examples of segments

Segments are sometimes the same across different projects, but every project is unique. - By product - By region - By store - By demographics - By marketing channel - Combinations of multiple other segments

Repeatability and scalability of analysis

Because segments are so important to the analysis, they deserve tooling for speed and consistency.

Segmentation should be stored as a first class citizen so that the entire level set can be exported with any desired segmentation. Segments are going to come up again and again, so important to get this one right. - We had a project titled Linnaeus and in the v1 version it would output cohort triangles of the overall rolled up numbers. It was somewhat valuable to see the cohort right away, but we also needed to be able to segment by any possible datapoint and export everything into excel.

The concept outlined in this idea is one way to possibly achieve this goal: 108.40.20.30 Data Analysis - An idea for raw data mapping and validation

Stratified sampling

This is a note to show correlation to another mathematical concept. 108.30.10.10.10 Data Analysis - Stratified sampling