Docs are not available for mobile use. Please use a desktop computer to view the documentation.

Boundary Detections

Cross-signal boundary detection results and multi-signal analysis

Purpose

The boundary_detections table stores the results of cross-signal boundary detection analysis. Unlike signal-specific boundaries, these detections identify periods where multiple signals collectively indicate significant transitions or state changes in user behavior or context.

Schema Definition

Field	Type	Description	Constraints
`id`	uuid	Unique detection identifier	Primary key, auto-generated
`userId`	uuid	Detection owner	Foreign key to users.id
`startTime`	timestamptz	Detection period start	Not null
`endTime`	timestamptz	Detection period end	Not null
`confidence`	real	Detection confidence (0-1)	Not null
`contributingSources`	text[]	Sources that contributed to detection	Not null
`detectionMethod`	detection_method	Method used for detection	Not null
`boundaryScope`	boundary_scope	Scope of boundary detection	Default: ‘all_signals’
`sourceSignal`	varchar(255)	Primary signal for single-signal scope	Optional
`createdAt`	timestamptz	Record creation timestamp	Auto-generated

Relationships

References:

users.id via userId - Detection owner

Indexes

idx_boundary_user_time - B-tree index on (userId, startTime, endTime) for temporal queries
idx_boundary_scope_signal - B-tree index on (boundaryScope, sourceSignal) for scope filtering

Enumeration Values

Detection Method

episodic_rule - Rule-based episodic event detection
ambient_change - Statistical change point detection
merged - Combination of multiple methods
speed_threshold - Rate-of-change based detection
gap_fill - Data continuity gap filling

Boundary Scope

single_signal - Detection within a single signal type
multi_signal - Detection across multiple related signals
all_signals - Detection across all available signals

Usage Examples

Get recent cross-signal detections

SELECT startTime, endTime, confidence, contributingSources, detectionMethod
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND startTime >= NOW() - INTERVAL '7 days'
ORDER BY confidence DESC, startTime DESC;

Find detections by scope

SELECT 
  boundaryScope,
  sourceSignal,
  startTime,
  endTime,
  confidence,
  array_length(contributingSources, 1) as source_count
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND boundaryScope = 'multi_signal'
ORDER BY source_count DESC, confidence DESC;

Analyze detection methods effectiveness

SELECT 
  detectionMethod,
  COUNT(*) as detection_count,
  AVG(confidence) as avg_confidence,
  AVG(EXTRACT(EPOCH FROM (endTime - startTime))/60) as avg_duration_minutes
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND startTime >= NOW() - INTERVAL '30 days'
GROUP BY detectionMethod
ORDER BY avg_confidence DESC;

Find high-confidence multi-source detections

SELECT 
  startTime,
  endTime,
  confidence,
  contributingSources,
  array_length(contributingSources, 1) as source_count
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND confidence >= 0.8
  AND array_length(contributingSources, 1) >= 3
ORDER BY confidence DESC, source_count DESC;

Get detections with specific contributing sources

SELECT startTime, endTime, confidence, contributingSources
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND 'fitbit' = ANY(contributingSources)
  AND 'location' = ANY(contributingSources)
ORDER BY startTime DESC;

Timeline analysis of detections

SELECT 
  DATE_TRUNC('day', startTime) as day,
  COUNT(*) as detection_count,
  AVG(confidence) as avg_confidence,
  mode() WITHIN GROUP (ORDER BY detectionMethod) as most_common_method
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND startTime >= NOW() - INTERVAL '30 days'
GROUP BY day
ORDER BY day;

Find single-signal focused detections

SELECT 
  sourceSignal,
  startTime,
  endTime,
  confidence,
  detectionMethod
FROM boundary_detections 
WHERE userId = 'user-uuid'
  AND boundaryScope = 'single_signal'
  AND sourceSignal IS NOT NULL
ORDER BY sourceSignal, startTime;