Part 4: Downloading the Dataset

4. The Flood Dataset

For this project, we utilize a custom compiled dataset containing historical weather records and observed flood labels.

Place the downloaded flood_prediction.csv file inside your project’s data folder. The final path must be: floodprediction/data/flood_prediction.csv

4.1 Dataset Columns Explained

The raw fields inside flood_prediction.csv are structured as follows:

• Climate Indicators:
  • Temp: Average regional temperature in Celsius (°C). Hotter temperatures can increase evaporation and atmospheric moisture capacity.
  • Humidity: Relative humidity percentage (%). Higher humidity indicates saturated air, making rainfall more imminent.
  • Cloud Cover: Cloud cover percentage (%). Represents atmospheric water vapor presence.
• Precipitation Metrics:
  • ANNUAL: Total annual rainfall in millimeters (mm).
  • Jan-Feb: Winter precipitation total in mm.
  • Mar-May: Summer precipitation total in mm.
  • Jun-Sep: Monsoon season precipitation total in mm. This is typically the most dangerous season for flooding.
  • Post-Monsoon & Averages:
    • Oct-Dec: Post-monsoon season precipitation total in mm.
    • avgjune: Average rainfall observed during the peak month of June in mm.
• Geographical & Classification Targets:
  • sub: Regional sub-basin precipitation index.
  • flood: Binary target column (1 = Flood, 0 = No Flood). This is the value our model must learn to predict.