A Machine Learning Approach to Optimize the Water Consumption for Irrigation of Rice Crop

In this paper a machine learning approach is proposed to provide required water for the rice crop. The precipitation level has been predicted using weather data for 13 years. The results of Linear Regressor, Multiple Regressor, and XGBoost have been compared. XGBoost Algorithm had been applied in Machine Learning to predict it using weather parameters on which rainfall is reliant, i.e., temperature, humidity, cloud cover, solar radiation, wind speed, and dew point. The experiment had been carried out twice by the split of ratio 70:30 and 80:20. The experiment had given accuracy of 91.89% and 92% respectively. This prediction had been used along with the Evapotranspiration rate and Crop factor to predict the crop water need as per the type of crop. This experiment showed a rousing result and utilized only an average of 2.71% water of the total requirement from external sources for the growth of the crop. This ensures that a passable amount of water is provided to the rice crop as per its requirement. This model can be used for judicious consumption of water for growth of crops which can help in meeting the future water demands of society.