Noise-aware ML pipeline for large-scale agricultural yield prediction using PySpark and LightGBM, with feature and label noise simulation, mitigation, and distributed training.

An interactive Q&A system that extracts actionable insights from agricultural data (1997–2014). It combines efficient EDA, a rule-based query engine, and a Gradio interface to answer questions on crop yields, rainfall, and production. Built in Jupyter Notebook with a modular, extensible pipeline.