Loading data...
⚠️ Experimental Analytics - These metrics are under development and may not be accurate
Mobility Entropy ?
--
Shannon entropy of location visits
What is Mobility Entropy?

Measures diversity in your location patterns. Higher = more varied/unpredictable movement. Lower = more routine behavior.

0-1 Highly routine (one dominant location)
1-2 Balanced (2-4 regular locations)
2+ Diverse, unpredictable movement

Example: Visiting 3 places equally = 1.58 bits. Visiting one place 90% of the time = ~0.5 bits.

Away Dwell Time
--
Average time spent at non-home locations per visit
Total Distance
--
Estimated distance traveled
Active Days
--
Days with location data

Home Detection Spatial

Auto-detected home clusters from nighttime (22:00-06:00) location density

Daily Exploration Radius Spatial

Maximum distance from nearest home each day (km)

Home vs Away Time Spatial

Hours spent at home (<500m from any detected home) vs away by day

Exploration Consistency Spatial

Standard deviation of daily radius - high variance = exploratory, low = routine

--
km std deviation
--

Radius of Gyration Over Time Spatial

Daily spatial spread from your activity centroid. Filter by transport mode to analyze movement patterns. Excludes travel days (>100km daily spread).

Dwell Time Distribution Spatial

Distribution of visit durations at non-home places

Dwell Time Statistics Spatial

Statistical measures of time spent at non-home locations

Median Dwell Time --
25th Percentile --
75th Percentile --
Unique Places Visited --
Total Away Visits --

Walk Score Analytics Walkability

Contextualizing walking behavior against neighborhood walkability scores (excludes home locations)

ℹ️ How this works
Walk Score (0-100) measures how walkable an area is based on proximity to amenities.
• 90-100: Walker's Paradise • 70-89: Very Walkable • 50-69: Somewhat Walkable • 0-49: Car-Dependent

Walking Rate calculation: GPS points are grouped into ~200m grid cells. Each point is matched to the nearest motion sensor reading (within 90 seconds). Walking Rate = (walking points / total points) × 100%. Example: 10 points in a cell (3 walking, 5 stationary, 2 unknown) → 30% walking rate. "Unknown" motion states count as non-walking.

Efficiency Index = (Your Walking %) ÷ (Expected Walking %). Expected rate assumes Walk Score 100 → ~70% walking max, scaling linearly.

All metrics exclude locations within 500m of detected home(s) to show true "out and about" behavior.
Loading Walk Score data...

Walking Efficiency Walkability

Are you walking as much as your environment enables? ⓘ

--
efficiency index
1.0 = walking as expected for walkability

Time by Walkability Walkability

Hours spent in areas by Walk Score category (away from home) ⓘ

Walking Rate vs Walkability Walkability

Your walking % at each location vs its Walk Score ⓘ

Walkability Statistics Walkability

Summary across away-from-home locations

Avg Walk Score ⓘ --
Time-Weighted Avg ⓘ --
Walking in High (70+) ⓘ --
Walking in Low (<50) ⓘ --

Place Attachment Spatial

Personally significant places based on visit frequency, dwell duration, and temporal regularity

High-Attachment Places --
Avg Visits (Top Places) --
Rank Location Score Visits
Calculating...

Activity Inference Speed Analysis

Inferring activity types from GPS movement speed when motion sensors report "unknown"

--
Point Pairs
--
Agreement Rate
--
Unknown → Classified
--
Walking Distance
ℹ️ How this works
Speed Calculation: Distance between consecutive GPS points ÷ time elapsed.
Uses Haversine formula for accurate spherical distance.

Speed Thresholds (research-based):
• Stationary: <0.15 m/s (0.5 km/h)
• Walking: 0.15-2.0 m/s (0.5-7.2 km/h)
• Cycling: 2.0-8.0 m/s (7.2-29 km/h)
• Driving: >8.0 m/s (29+ km/h)

Likely Walking Routes: Consecutive segments where speed stays in walking range for 3+ points with directional movement (not just stationary).
Analyzing GPS trajectories...

Inferred Activity Speed

Speed-based activity classification ⓘ

Sensor vs Inferred Speed

How well does speed match sensor data? ⓘ

Speed Distribution Speed

Distribution of movement speeds ⓘ

Likely Walking Routes Speed

Detected walking-speed segments

Walking Segments ⓘ --
Total Walking Distance ⓘ --
Avg Walking Speed ⓘ --
Unknown Reclassified ⓘ --

Transportation Mode Detection Comparison Motion Classification

Comparing 4 algorithms for differentiating transportation modes (walking, cycling, bus, car, train) using GPS data

--
Best Algorithm
--
Overall Accuracy
--
Train/Bus Detection
--
Processing Time
ℹ️ Methodology
Algorithm 1 - Baseline: Simple speed thresholds (walk <2 m/s, cycle <8 m/s, bus <15 m/s, car <25 m/s, train >25 m/s)

Algorithm 2 - Percentile95: Uses 95th percentile speed in sliding window to handle speed variance

Algorithm 3 - Stop Pattern: Analyzes frequency of stops (bus = frequent stops, train = few stops)

Algorithm 4 - Heading Change: Analyzes direction changes (train = straight, bus = more turns)

Ground Truth: Derived from motion sensors + speed-based inference for automotive mode distinction
Analyzing transportation patterns...

Algorithm Performance Comparison Motion

Performance metrics for each algorithm ⓘ

Algorithm Description Overall Accuracy Train Detection Bus Detection Car Detection Processing Time
Loading...

Baseline Confusion Matrix Motion

Simple speed threshold classification ⓘ

Percentile95 Confusion Matrix Motion

95th percentile speed analysis ⓘ

Stop Pattern Confusion Matrix Motion

Stop frequency analysis ⓘ

Heading Change Confusion Matrix Motion

Direction change analysis ⓘ

Feature Importance Analysis Motion

Key factors for distinguishing modes ⓘ