Files
personal-tracker/scripts/ingest/geospatial_data_relationships.txt
2025-09-25 21:01:15 +01:00

185 lines
7.4 KiB
Plaintext

GEOSPATIAL DATA RELATIONSHIPS IN PERSONAL TRACKER
==================================================
This document explains how the geospatial datasets in the timeline_csv folder are interconnected
and structured to provide a complete picture of location-based personal tracking data.
OVERVIEW
--------
The location tracking system uses a hierarchical approach with semantic segments as the master
index that coordinates different types of location data. The data is organized into movement
periods (travel) and stationary periods (visits), creating a complete chronological timeline.
CORE DATASETS AND THEIR RELATIONSHIPS
=====================================
1. SEMANTIC_SEGMENTS.CSV - The Master Index
--------------------------------------------
Purpose: Acts as the central orchestrator that defines time-based segments
Key Fields:
- segment_index: Unique identifier linking all other datasets
- startTime/endTime: Time boundaries for each segment
- has_visit: Boolean indicating if segment contains visit data
- has_timeline_path: Boolean indicating if segment contains movement data
This dataset defines the temporal structure and determines which other datasets contain
data for each time period.
2. TIMELINE_PATH_POINTS.CSV - Movement Data
--------------------------------------------
Purpose: GPS tracking data during travel/movement periods
Key Fields:
- segment_index: Links to semantic_segments
- point_index: Order of GPS points within a segment
- time: Precise timestamp for each GPS reading
- lat/lon: Geographic coordinates
- raw_point: Original coordinate string
Relationship: Contains data ONLY for segments where has_timeline_path=1 in semantic_segments.
These represent periods when the person was moving between locations.
3. VISITS.CSV - Stationary Location Data
-----------------------------------------
Purpose: Information about places where the person stayed for extended periods
Key Fields:
- segment_index: Links to semantic_segments
- top_place_id: Google Places API identifier
- top_semantic_type: Category (HOME, WORK, UNKNOWN, etc.)
- top_lat/top_lon: Geographic coordinates of the visit location
- startTime/endTime: Duration of the visit
- visit_probability: Confidence that this was actually a visit
Relationship: Contains data ONLY for segments where has_visit=1 in semantic_segments.
These represent periods when the person was stationary at a specific location.
4. FREQUENT_PLACES.CSV - Location Reference Data
------------------------------------------------
Purpose: Registry of commonly visited locations with semantic labels
Key Fields:
- placeId: Google Places API identifier (links to visits.top_place_id)
- label: Semantic meaning (HOME, WORK, or empty for unlabeled)
- lat/lon: Geographic coordinates
Relationship: Acts as a lookup table for visits.csv. The placeId field provides
cross-references to identify and categorize frequently visited locations.
5. RAW_SIGNALS.CSV - Raw GPS Data
---------------------------------
Purpose: Unprocessed GPS signals from the device
Key Fields:
- raw_index: Sequential identifier
- timestamp: When the GPS signal was recorded
- lat/lon: Geographic coordinates
- accuracyMeters: GPS accuracy measurement
- altitudeMeters: Elevation data
- speedMetersPerSecond: Movement speed
- source: Data source type
Relationship: This is the foundation data that gets processed into timeline_path_points
and visits. It represents the raw GPS signals before semantic interpretation.
SUPPORTING DATASETS
===================
6. FREQUENT_TRIPS.CSV - Trip Pattern Analysis
----------------------------------------------
Purpose: Analysis of regular travel patterns (like commutes)
Key Fields:
- trip_index: Unique identifier for trip patterns
- startTimeMinutes/endTimeMinutes: Time of day patterns
- durationMinutes: Typical trip duration
- commuteDirection: HOME_TO_WORK or WORK_TO_HOME
- waypoint_count: Number of stops in the trip
7. FREQUENT_TRIP_WAYPOINTS.CSV - Trip Waypoint Details
------------------------------------------------------
Purpose: Specific locations that are part of frequent trips
Key Fields:
- trip_index: Links to frequent_trips.csv
- waypoint_order: Sequence of stops in the trip
- waypoint_id: Links to frequent_places.placeId
8. FREQUENT_TRIP_MODE_DISTRIBUTION.CSV - Transportation Analysis
---------------------------------------------------------------
Purpose: Analysis of transportation methods used
Key Fields:
- trip_index: Links to frequent_trips.csv
- mode: Transportation type (WALKING, DRIVING, etc.)
- percentage: How often this mode was used for this trip
9. TRAVEL_MODE_AFFINITIES.CSV - User Preferences
------------------------------------------------
Purpose: User's preferred transportation methods
Key Fields:
- mode: Transportation type
- affinity: Preference score
DATA FLOW AND RELATIONSHIPS
============================
1. RAW COLLECTION:
raw_signals.csv contains all GPS pings from the device
2. TEMPORAL SEGMENTATION:
semantic_segments.csv divides time into logical periods based on movement patterns
3. MOVEMENT vs. STATIONARY CLASSIFICATION:
- Movement periods → timeline_path_points.csv (detailed GPS tracking)
- Stationary periods → visits.csv (location identification and categorization)
4. LOCATION IDENTIFICATION:
frequent_places.csv provides semantic meaning to visited locations
5. PATTERN ANALYSIS:
frequent_trips.csv, frequent_trip_waypoints.csv, and frequent_trip_mode_distribution.csv
analyze regular patterns and transportation preferences
EXAMPLE DATA FLOW
==================
Segment 0 (Movement): 2013-12-31 22:00 - 2014-01-01 00:00
- semantic_segments: has_timeline_path=1, has_visit=0
- timeline_path_points: Contains GPS coordinates during this travel period
- visits: No data for this segment
Segment 1 (Visit): 2013-12-31 22:29 - 2014-01-01 17:10
- semantic_segments: has_timeline_path=0, has_visit=1
- timeline_path_points: No data for this segment
- visits: Shows visit to place ChIJyaJWtZVqdkgRZHVIi0HKLto (HOME)
- frequent_places: Confirms this placeId is labeled as "HOME"
QUERYING STRATEGIES
===================
To get complete journey information:
1. Query semantic_segments for time range
2. For movement segments: Join with timeline_path_points on segment_index
3. For visit segments: Join with visits on segment_index
4. Enhance visit data by joining visits.top_place_id with frequent_places.placeId
To analyze location patterns:
1. Use frequent_places for location categories
2. Use frequent_trips for commute patterns
3. Use travel_mode_affinities for transportation preferences
COORDINATE SYSTEMS
==================
All latitude/longitude data uses WGS84 decimal degrees:
- Latitude: Positive = North, Negative = South
- Longitude: Positive = East, Negative = West
- Precision: Typically 6-7 decimal places (meter-level accuracy)
TIME ZONES
==========
All timestamps include timezone information (typically +00:00 or +01:00 for UK data).
Time ranges in semantic_segments define the boundaries for linking other datasets.
DATA COMPLETENESS
=================
- Not all segments have both movement and visit data
- Some segments may have neither (gaps in tracking)
- Visit probability scores indicate confidence levels
- Missing coordinates in raw_signals are represented as empty fields
This hierarchical structure allows for both detailed movement tracking and high-level
pattern analysis while maintaining semantic meaning about the places visited.