Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

name: intents
description: The closed intent taxonomy: the kinds of question an analysis answers, how each sounds, and the field shapes it fits. Use to map a user's question to an intent ID.
type: reference
kind: reference

Intents

The closed set of question kinds an analysis can answer. Operator purposes and examples cite these IDs; a shape lists the roles a cohort’s fields must fill for the question to be answerable.

Analytic intents

benchmark

Benchmark against a reference.

Sounds like:

  • How does this compare to the total?
  • Is this group above or below the population?
  • How does this wave compare to the last?

Shapes:

  • measure (numeric|categorical|bool, 1 or more) + group (categorical|bool|date, exactly 1)

change_over_time

Change over time.

Sounds like:

  • Is it going up?
  • How does this month compare to last?

Shapes:

  • time (date, exactly 1) + measure (numeric|categorical|bool, 1 or more)

compare_groups

Compare groups.

Sounds like:

  • Do A and B differ?
  • Is this segment different from the rest?

Shapes:

  • outcome (numeric|categorical|bool, exactly 1) + group (categorical|bool, exactly 1)

composition

Mix and share.

Sounds like:

  • What’s the mix?
  • What share does each option have?
  • Which attributes go with which brand?

Shapes:

  • category (categorical|bool|set, exactly 1) + by (categorical|bool, optional)

data_quality

Data quality.

Sounds like:

  • Is this data trustworthy?
  • Who answered carelessly?
  • How much is missing?

Shapes:

  • fields (numeric|categorical|date|bool|set, 1 or more)

describe

Describe a measure.

Sounds like:

  • What’s the typical value of X?
  • What’s the total X?
  • How spread out is X?

Shapes:

  • measure (numeric, 1 or more)
  • measure (categorical|bool|set, 1 or more)

distribution_shape

Distribution shape.

Sounds like:

  • Is it normally distributed?
  • Are there outliers?
  • Is it skewed?

Shapes:

  • measure (numeric, 1 or more)

drivers

Find drivers.

Sounds like:

  • What explains Y?
  • Which factors matter most for Y?

Shapes:

  • outcome (numeric|categorical|bool, exactly 1) + predictors (numeric|categorical|bool, 1 or more)

flows

Flows between states.

Sounds like:

  • Where do customers move between states?
  • Who switched from A to B?

Shapes:

  • from (categorical|bool, exactly 1) + to (categorical|bool, exactly 1)

measure_construct

Measure a construct.

Sounds like:

  • Do these questions measure one thing?
  • Combine these items into a score.

Shapes:

  • items (numeric, 2 or more)

relationship

Relationship between measures.

Sounds like:

  • Do X and Y move together?
  • Is X associated with Y?

Shapes:

  • measures (numeric, 2 or more)
  • categories (categorical|bool, exactly 2)

segment

Segment into groups.

Sounds like:

  • Are there natural groups of customers?
  • Split this into tiers.

Shapes:

  • measures (numeric, 1 or more)

Tooling intents

lookup

Look up records.

Sounds like:

  • Show me the record for this ID.
  • Fetch the rows for this key.

Shapes:

  • key (numeric|categorical|date, 1 or more)

prepare

Prepare data.

Sounds like:

  • Keep only last year’s rows.
  • Derive a new column from these fields.

Shapes:

  • fields (numeric|categorical|date|bool|set, 1 or more)

simulate

Simulate data.

Sounds like:

  • Generate a synthetic copy of this cohort.
  • Make test data with this schema.

Shapes:

  • fields (numeric|categorical|date|bool|set, 1 or more)