Skip to content
GPS Staypoint Assignment
Inbox
Initializing search
gps-assignment-reports
Home
Map
Threads
Progress
Notes
Analysis
Inbox
GPS Staypoint Assignment
gps-assignment-reports
Home
Map
Threads
Threads
2026
2026
August
August
thesis summary
problem and positioning
outlook
ground truth data
error analysis
complexity and transfer
calibration
assignment framework
Progress
Progress
2026
2026
August
August
phd interview slides (pdf)
July
July
does downsampling matter?
rq2 diagnostics and rq3 ceiling check
rq1 rebuilt with grouped cv
ml baseline
mnl model development
Notes
Notes
2026
2026
August
August
glossary
July
July
study area
spatial entity layers
research questions
overfitting signature
motivation
limitations
ground truth dataset
future work
framework overview
f_temporal
f_distribution
f_distance
f_directional
experiment design
downsampling to miles density
contributions
candidate selection
tension with the street-space argument
self-report form
positioning against deterministic assignment
extending map matching to stationary stays
ground truth cleaning
group e: ambiguous edge cases
group d: movement type
group c: indoor position
group b: sky view obstruction
group a: vertical position
candidate density as boundary-matching difficulty
grouped cross-validation
sigma
share_type
multicollinearity is structural
model evaluation metrics
Analysis
Analysis
2026
2026
July
July
raw-density refit
raw-density alpha profile
downsampling shrinks the candidate search area
sigma in tree importance
ml ceiling check, row level
fold difficulty
ground truth drop list
confusion matrix
candidate density bias toward building
stepwise vif elimination
model progression
model 6, re-derived at calibrated alpha
model 5: alternative-specific constants (rejected)
model 4a: alternative-specific parameters
model 2: share-type and sigma
model 1: base
interaction term screen (α = 1)
final model
final model correlation matrix
the f_distance² artifact
f_directional: shared vs split
directional scope ablation (3a vs 3b)
alpha calibration
ml baseline, stay level (superseded)
r_buffer: 10 m vs 15 m
Inbox
Inbox
Inbox
Open items not yet settled with sensei.
Nothing published here yet.
Back to top