SPAS-Dataset-BD: Dataset for smart precision agriculture system in Bangladesh

Area of Science:

• Agricultural Science
• Data Science
• Environmental Science

Background:

• Precision agriculture relies on data, but low-income countries like Bangladesh lack region-specific datasets.
• Existing data often fails to capture local agroclimatic conditions and farming practices.

Purpose of the Study:

• To introduce SPAS-Dataset-BD, a comprehensive dataset for precision agriculture in Bangladesh.
• To provide a robust, localized data resource for research and policy-making in South Asian agriculture.

Main Methods:

• A hybrid approach combining secondary data from the Bangladesh Bureau of Statistics (BBS) 2022 Yearbook and primary on-field surveys.
• Data collection involved 223 farmers across ten districts, covering 73 crop types and 12 features.
• Ensured data robustness through missing-value handling, deduplication, and cross-validation against official statistics.

Main Results:

• SPAS-Dataset-BD contains 4191 records with detailed agronomic and environmental features.
• Demonstrated dataset robustness and suitability for machine learning applications.
• Highlighted potential uses in 73-class crop classification, yield forecasting, and IoT-based irrigation scheduling.

Conclusions:

• SPAS-Dataset-BD is a valuable, large-scale, and methodologically transparent resource for precision agriculture in South Asia.
• The dataset's contextual richness supports localized agricultural research and informed policy decisions.
• Facilitates advancements in data-driven farming practices and resource optimization.