A One-Year Context-Aware Synthetic IoT Dataset for Smart Waste Management in Dhaka, Bangladesh: 43.8 Million Hourly Observations
Description
This dataset provides a one-year, context-aware synthetic IoT benchmark for smart waste-management research in Dhaka, Bangladesh. It represents 5,000 persistent synthetic smart-bin digital twins simulated hourly throughout 2025, producing exactly 43,800,000 observations. The simulation incorporates land-use and population context, institutional and market activity, waste composition, informal recovery, weather seasonality, Ramadan and Eid scenarios, road accessibility, collection and overflow dynamics, and IoT sensor behaviour. The dataset contains 59 hourly observation variables and 29 persistent bin-level variables. Geographic coordinates are synthetic points generated around representative Dhaka neighborhood centers and do not represent actual municipal smart-bin locations. The primary archival release is provided in Parquet format as 92 observation files grouped into 23 ZIP archives, together forming one logical dataset, with persistent bin metadata provided in bins.parquet. An equivalent CSV/GZIP mirror containing 44 observation parts and bins.csv.gz is also provided through the project data repository.
Files
Steps to reproduce
1. Install Python 3 and the required packages using: pip install -r requirements.txt 2. Generate the complete synthetic dataset: python generate_v4_0.py 3. Validate the generated dataset: python validate_v4_0.py 4. For the Parquet release, validate the exported partitions using: python validate_export_parts.py The generator uses a fixed NumPy random seed of 20260902. The expected output contains 5,000 persistent synthetic bins, 8,760 hourly observations per bin, and exactly 43,800,000 hourly observations for the year 2025. The public release contains 92 Parquet observation partitions together with bins.parquet. Detailed generation methodology, variable definitions, calibration information, and validation procedures are included in the documentation package.