An interactive map of criminal activity (thefts, robberies, frauds) in Lviv, Ukraine, built from the city's open data portal — a side project exploring whether public open data could answer a very practical question: which neighborhoods are safer to live in?
Live demo: https://anatolis.shinyapps.io/ShinyLeaflet/
Over 13,000 police-registered cases from 2015, sourced from the Lviv Open Data Portal. Each record's street address was geocoded to coordinates via the OpenStreetMap and Google Maps geocoding APIs (geo.py), then aggregated up to Lviv's district boundaries (extracted from Wikimapia, district.geo.json) to compute a per-district crime rate normalized by area and population.
The analysis (see the notebooks below) went through a few iterations:
- Plot every case as a dot on the map — turned out too noisy to reveal any pattern.
- Heatmap — better, but still hard to read at city scale.
- District choropleth — settled on this: districts colored on a green→red gradient by crime rate, which is the clearest way to compare areas at a glance.
| Raw case dots | Cases by police department |
|---|---|
 |
 |
Data manipulations.ipynb/Data manipulations 2.ipynb— fetching, cleaning, and geocoding the raw datasetsLviv districts.ipynb— building district boundary polygonsInitial EDA.ipynb— first exploratory plots and color scheme choicesConclusion.ipynb— write-up of findings and design decisions
ShinyLeaflet/ contains the R Shiny + Leaflet app that serves the interactive map, with toggles for case type (theft/robbery/fraud) and view (dots/heatmap/by-district).
A university side project from 2017 — data is a one-year snapshot from 2015 and the app isn't actively maintained, but the live demo above still shows the visualization in action.