Tartu Smart Bike · July 2019

Interactive Findings & Visual Gallery

Explore the key results from temporal, spatial, statistical, machine learning, and network analyses of the Tartu Smart Bike system. All assets are generated from the open dataset and live in this repository.

View on GitHub Read full EDA report (Markdown)

Time Series & Forecasting

Seasonal Decomposition

Hourly demand · trend / seasonal / residual
Seasonal decomposition

Forecast (SARIMA)

1-week ahead demand prediction
Hourly pattern

Statistical Insights

User Segmentation

3 usage segments
User segmentation

Community Detection

Louvain communities
Community distribution

Centrality Metrics

NetworkX centralities
Centrality metrics

Station Network

73 stations · 3,520 routes
Station network

Machine Learning

User Behavior Clustering

K-means clusters
User behavior clustering

Anomaly Detection

Isolation Forest results
Anomaly detection

Interactive Maps & Charts

GPS Heatmap

Folium · zoom & pan

How to reproduce

Local reproduction

Install dependencies and run:

pip install -r requirements.txt
python scripts/01_data_preprocessing.py
python scripts/02_run_eda.py

Streamlit dashboard

streamlit run dashboard.py
Built from open Tartu Smart Bike data (July 2019). All visualizations and HTML assets are generated by the analysis scripts in this repository.