Dataset and Source Code for: Maximizing Cotton Returns: A Deep Learning Approach to Weather Integrated Price Forecasting in Tamil Nadu

This dataset contains the complete data and source code supporting the study on cotton price forecasting using deep learning in Tamil Nadu, India. It includes: (1) weekly cotton price data for Salem and Perambalur markets (2010–2025) from AGMARKNET, integrated with meteorological variables (rainfall, temperature, solar radiation, humidity) sourced from the NASA POWER database; (2) cleaned and preprocessed datasets for both districts; and (3) Jupyter Notebook source code implementing the hybrid GRU–LSTM deep learning model and comparative neural network architectures for univariate and multivariate price forecasting. The dataset also supports Benefit–Cost Ratio analysis of cotton-based cropping systems in Tamil Nadu.本数据集涵盖支撑印度泰米尔纳德邦（Tamil Nadu）基于深度学习开展棉花价格预测研究的完整实验数据与配套源代码，具体包含以下内容： (1) 2010-2025年源自AGMARKNET的塞勒姆（Salem）与佩兰巴卢尔（Perambalur）两地农产品批发市场的周度棉花价格数据，配套整合了美国国家航空航天局POWER数据库（NASA POWER）获取的气象变量数据，涵盖降雨量、气温、太阳辐射与空气湿度； (2) 针对上述两个地区的经清洗与预处理的数据集； (3) 实现混合门控循环单元-长短期记忆网络（GRU-LSTM）深度学习模型，以及用于单变量与多变量价格预测的对比性神经网络架构的Jupyter Notebook源代码。 本数据集亦可用于支撑泰米尔纳德邦棉作种植系统的效益成本比（Benefit–Cost Ratio）分析。