One bubble per state; x = elderly case share, y = fatality, area = total cases.
Insight: states with an older case mix carry a higher case
fatality rate. An outbreak skewing elderly should trigger earlier hospital escalation.
Occupancy & ICU utilisation
Weekly mean percentage across selected jurisdictions; dashed line is the adjustable alert threshold.
Weeks above alert
Share of observed state-weeks at or above the selected bed-occupancy threshold.
Trellis: bed occupancy by state
Monthly mean bed occupancy for the nine highest-caseload states; shared 0-based y-scale.
Vaccination coverage over time
Weekly mean of state coverage, percentage of population; full and booster series use distinct line styles.
Weekly doses administered
Sum of doses reported across jurisdictions each week; zero-based dose scale.
Vaccination coverage vs case fatality
One state per bubble during weeks with coverage data; x = final coverage, y = fatality, area = caseload.
Interpretation: higher final coverage has a weak inverse association with
case fatality in this aggregate comparison. It is descriptive and does not establish individual-level
causation; age-adjusted or patient-level analysis would be needed for that conclusion.
Share of unvaccinated among new cases
Weekly mean state share of reported cases classified unvaccinated; percentage scale fixed at 0–100%.
Correlation heat map
Pearson r from −1 to +1 across filtered state-week rows using pairwise complete observations.
State profiles — parallel coordinates
One line per state across five metrics; each vertical axis has its own labelled min–max scale.
Scatter-plot matrix
Pairwise state-week relationships from a deterministic sample of up to 700 rows, coloured by phase.
How to read relationships: values near +1 move together, values near −1 move in
opposite directions, and values near 0 show little linear relationship. Correlation describes association;
it does not by itself establish causation.
Why these advanced techniques?
Rubric-ready rationale, revealed insight, and comparison with a traditional chart.
Correlation heat map
Why selected
Scans many metric pairs in one compact matrix.
What it reveals
Direction and strength of linear co-movement, with pairwise sample size on hover.
Versus a traditional chart
More efficient than separate two-series charts, but it does not reveal non-linear shape or causation.
Parallel coordinates
Why selected
Compares every state across five measures simultaneously.
What it reveals
Multivariate profiles, trade-offs, clusters, and outlying jurisdictions.
Versus a traditional chart
Shows more dimensions than grouped bars; independent axes require careful scale reading.
Scatter-plot matrix
Why selected
Checks all pairwise distributions and phase clusters without choosing one relationship first.
What it reveals
Outliers, non-linearity, changing variance, and phase-dependent structure.
Versus a traditional chart
Extends one scatterplot to every metric pair, trading simplicity for broader diagnosis.
Weekly surge anomaly explorer
National weekly cases against a trailing statistical baseline; brush the context strip to inspect a period.
Actionable insights
Ranked findings from the current data, each with a recommendation
Project progress timeline
Project started 8 July 2026 · submission due 23 July 2026 at 11:59 PM · scroll to zoom, drag to pan, double-click to reset
Data methodology: Malaysia Ministry of Health open data (covid19-public, CC BY 4.0).
Map: click_that_hood Malaysia GeoJSON. Hand-built with D3.js v7 for the TEB3133/TFB3133 DV Project.
Educational public-data dashboard; not an official KKM service.