Parameter Optimized Random Forest Method for Crop Yield Prediction

Abstract

     Agriculture is critical to our country's economic development. Agriculture fields are deteriorating these days due to a variety of obstacles and issues such as temperature, rainfall, and so on. As a result, a prediction model is required to assist farmers in making better decisions on crop yield forecast. Machine learning is a critical decision-making tool in a variety of applications, including crop yield forecasts to anticipate which crops should be planted. Several machine learning techniques for agricultural yield prediction have been proposed in the literature, including Random Forest, Support Vector Machine, and Regression. However, there is still developments of technology. Within the study's framework, we have proposed Parameter Optimized Random Forest Classifier (PORF) method for crop prediction to recommend suitable crop for farmers better cultivation. In this proposed method optimization technique is implemented to tune the parameter that fit to the model and feature importance is calculate for the features thus the less important features can be dropped for the better accuracy. The proposed model is tested using various real-world datasets for performance benchmark. The proposed PORF method outperforms with traditional classifiers in terms of prediction accuracy.