Trace
SIGMA Lab
Investigate the pattern behind the signal.
Loading formal observations…
Spatial Intelligence & Geographic Modeling for Action
Trace
Investigate the pattern behind the signal.
Loading formal observations…
Trace
Investigate the pattern behind the signal.
Evidence
FormalFormal observations · OpenDengue v1.3
Formal observed data
This is reported historical disease data, not model output.
DenguePhilippines · nationalNational / Weekly2012–2023OpenDengue v1.3 · Temporal extract (WPRO) · CC BY 4.0
Earliest observation
Dec 30, 2012
Week start
Latest observation
Nov 25, 2023
Week end
Weekly observations
544
of 569 weeks in range
Resolution
National · weekly
No sub-national formal series used
Summary statistics
Weekly figures use the 544 recorded weeks only; missing weeks are excluded, not counted as zero.
Mean weekly count
3,513.6
Median weekly count
2,537.5
Maximum weekly count
21,743
Peak week of Jul 28, 2019
Minimum weekly count
132
Week of Jul 2, 2023
Complete epi years
3
2013, 2019, 2020 · every week recorded
Highest complete-year total
437,777
2019 · sum of 52 weeks
Annual totals are shown only for epi years with every week recorded; other years have partial sums, labelled as such.
Data quality / coverage
Partial coverage
Data quality note
Reported dengue cases per week
Philippines, national · Dec 30, 2012 – Nov 25, 2023 · drag the handles below to zoom
47 of 52 weeks recorded · record ends within this year; no annual total
* Incomplete epi year. Shaded columns are weeks with no record.
Seasonal profile by epidemiological week
Median and min–max across epi years with a record for that week (weekly rows only). Descriptive, not a baseline. Week 53 exists only in 53-week years.
Historical baseline
ModelBaseline: Not yet fitted
SIGMA will compare current observations against historical expected activity before producing anomaly indicators.
Model not yet fitted
AnomalyResultrecords (observed, expected, range, score, classification, model and version) and this panel draws observed vs expected vs alert threshold.Level 1 · Not started
Seasonal / historical baseline
Expected weekly activity from prior years at the same point in the season.
Level 2 · Not started
Farrington / EARS-style anomaly detection
Excess over the baseline with an explicit alert threshold.
Level 3 · Not started
Forecasting baseline
SARIMA or Holt-Winters, backtested with rolling origins.
Level 4 · Not started
ML benchmark
XGBoost or comparable, compared against levels 1–3.
Level 5 · Not started
Spatiotemporal model
Requires reconciled sub-national formal data.
Level 6 · Not started
Scenario simulation
Assumption-driven what-ifs; never presented as forecasts.
Provenance
Describe what you want to see. This first phase does not query a live backend — it previews how a future natural-language-to-query workflow could work.