Reanalyze it starting from here, but apply spatial and temporal analysis techniques. MacroCounty and MicroCounty calculates nearest neighbor based on county, while MacroGrid and MicroGrid is based on Grid we generated. Models folder include not only Mapper Algorithm from TDA, but also three machine learning methods to identify hotspot we generated from Getis-ord.

1. Youbike (deprecated) (New Taipei, Taipei, Taichung)
2. Parking Lot
3. Metro Rapid Transit: Transport Data Exchange(捷運站出入口基本資料)
4. Youbike: 自行車租借站位資料
5. Taiwan shape and 鄉鎮市區界線(TWD97經緯度)1140318

1. Attractions
2. School Data

• Youbike, MRT, Attractions data sources installed in 2025/04/15
  
• Parkinglot data sources installed in 2025/04/21
  
• Taiwan shape 2025/06/06
• Traffic accident data installed in 2025/10/18, include data until Sep

conda create --name ST-RTA python=3.10
conda activate ST-RTA
pip install -r requirements.txt

• DataCombine: Combine A1 and A2 separately in each year, month
  
• DataPreprocess: Preprocess data for grid analyze grid_gi and model input data all_features
  
• Model: Run ML model from all_features
  
• FilterforMapper: Get the filter function for Mapper from all_features data
  
• GetCCforCombineddata: Add County name to combined data, get combineddata_with_CC

Some reference papers are stored in notion