Develop an Ai-Based Crop Predictor for Haryana Farmers

The AI-Based Crop Predictor for Haryana Farmers is a Machine Learning project designed to help Indian farmers, especially from Haryana, make data-driven decisions about which crop to grow next. Many farmers struggle to identify soil moisture and other scientific factors. This project uses ai and simple farmer inputs to predict best-suited crops for given conditions, without requiring any complex knowledge like npk values or soil ph. The objective is to predict soil moisture using historical weather and satellite data, and then use this predicted moisture with farmer-friendly inputs (district, date, temperature, rainfall, previous crop, soil type) to predict the best crop to grow. The goal is to simplify agricultural data science for real-world rural use cases. The workflow involves: Step 1: Moisture Prediction. This step includes cleaning and exploring the Haryana dataset (State, District, Date, Year, Month, Moisture, Source), performing null handling, dropping unnecessary columns, type conversions, and EDA (graphs, distributions, relationships). The model used is LightGBM Regressor (lgb.LGBMRegressor) with hyperparameter tuning via GridSearchCV, achieving a training accuracy of 99.75% and testing accuracy of 96.28%. A reusable moisture prediction function is defined for new farmer inputs. Step 2: Month Feature Engineering. This involves implementing sin/cos transformations on month features to capture the cyclical nature of months (e.g., January and December are close in time). Step 6: Best Crop Prediction Model. This final step combines all previous outputs into an AI model. Inputs used are District, Year, Month, Day, Temperature, Rainfall, Predicted Moisture (15cm), Previous Crop, and Soil Type. The output is a recommended best crop to grow.