Outlier Hunter Game

Mission control

Configure a round, investigate records, then submit your marked observations.

Game mode

Dataset scenario

Difficulty

Detection method

Primary chart

Feature scaling

Dataset size: 40

Outliers: 5

Noise: 25%

Sensitivity: 2.5

Neighbours: 5

Time limit: 90s

Random seed

Assistance

Allow hints

Sound effects

Outlier pattern

Missing-value treatment

Animation speed

Import custom CSV

The first three numeric columns become game features.

Accessibility

High contrast

Reduced motion

Larger text

Score

Time

90s

Marked

Streak

Precision

—

Recall

—

Canvas investigation field

Click points to mark or unmark observations. Hover to inspect record details.

Find unusual product sales records

Look for observations that differ from the main pattern.

Round 1

Yellow ring: marked Red pulse: revealed outlier Hover a point

Detection assistant

Interquartile rangeIQR

Flags values beyond quartile fences. It is robust against extreme values.

Hint ready.

Hints reveal investigation strategies and reduce your score.

Round active.

Mark suspicious observations, compare evidence, then submit.

Achievements

Data table investigation

Sort columns, search records, inspect values, and select rows.

Showing records.

Visual evidence dashboard

Zoom, hover, select points, and compare distribution evidence.

Round report and exports

Review metrics, preserve results, or replay the same seeded dataset.

True positives

False positives

False negatives

F1 score

—

Responsible investigation: an unusual record is not automatically wrong. Verify its source, context, units, collection process, and business meaning before correcting or removing it.