Place discovery and crowd forecasting
DuckIn
DuckIn is a mobile app that helps you understand how crowded a place may be before you go, and when it might be calmer. It combines place discovery with hourly crowd forecasts in Istanbul, Ankara and İzmir.
- Developer
- Mustafa Erhan Kaya
- Technology
- Flutter, Dart, TypeScript, PostgreSQL, PostGIS, Supabase, Mapbox
From a location to a time to visit
Knowing where a venue is does not tell you when to go. DuckIn combines historical visit patterns, weather, traffic, events and calendar signals to help people plan a visit.
Geographic data and the mobile experience
A Mapbox map in the Flutter app brings place discovery and forecasts into the same flow. PostgreSQL and PostGIS support nearby-place queries and geographic data. Scheduled collectors feed the backend.
Explainable forecasts
The TypeScript forecasting core takes explicit inputs and time. It produces current and hourly forecasts, confidence information and contributing factors. These are predictions built from several signals, rather than directly measured occupancy counts at every venue.
Comparing changes on the same data
Data collection is separated from scoring. Replaying captured inputs lets engine changes be compared against the same examples instead of a constantly changing live-data target.
More of my work
ZINK!
ZINK! is a mobile word game built around a clue and a letter count. Revealing letters costs points, making the decision to solve without help central to each round. It offers Turkish and English content, daily challenges and crossword puzzles.
Read the project storyPyralis
Pyralis is a mobile product that brings scattered gaming libraries, reviews and player history together. It connects digital games and tabletop experiences in a shared player profile.
Read the project story