ELISA CHALLENGE Synthetic source data
ELISA MOCK DATASET

Service quality by cohort

Compare throughput and latency by region, application category and radio type.

ANALYSIS SCOPE

Regional service-quality comparisons

Each cohort groups subscriber records by region, application category and radio type.

Service development
Release policyMinimum distinct subscribers per cohort
Source observations 1,099,340Full Elisa mock dataset
Protected cohorts 688/ 1,089401 small cohorts withheld
Observation coverage 99.6%On targetTarget ≥ 90% of source observations
Throughput error 0.8%MeasuredTarget ≤ 5% · released cohorts only

Highest cohort averages

Highest download throughput · subscriber-weighted means

REGION / CATEGORYSOURCE SCALE
CURRENT POLICY

Release safeguards

No subscriber identifiers, exact timestamps, base stations, or persistent tracking tokens in the release.

At least 50 subscribers per cohort
Missing values are excluded from averages
Each row includes its calculation method
Re-identification risks remain. See the privacy assessment.

Processing steps

Identifiers are removed before the cohort table is released.

01 Local source02 Remove & aggregate03 Suppress & round04 Release + receipt
Release 9f7e9b63 / Balanced · k ≥ 50Elisa challenge · AaltoAI 2026