Dataset browser

Tartu Safety dataset

Interactive table view sourced from tartu_risk_dataset_2.csv.

Back to forecasting map

Rows -

Districts -

Dataset window -

Source status Loading

Critical methodology explanation

This dataset is not neutral

The dataset was created through many subjective human choices. Those choices directly shape what the model sees, what it ignores, and therefore what outcomes it can produce.

We chose source materials using our own judgement and local knowledge about which references seemed useful enough to inform this prototype.

We defined the dataset features through subjective human decisions about which variables mattered and how they should be represented.

We used subjective local know-how to categorize each suburb for features such as foot_traffic, area_type, socially_vulnerable_zone, minorities_zone, student_zone, and nightlife_zone.

All of this human data work directly affects the outcomes. There is no neutrality here: small human interactions accumulated while this dataset was created.

Reference data provenance

This prototype dataset is synthetic, but it is partly informed by real public reference data from Tartu and police-related published materials.

Police callout reference

Incident pattern assumptions are partly grounded in published Väljakutsed material, where the stated public data source is the police tactical management database KILP.

Open official Väljakutsed source