Hybrid extreme learning machine based bidirectional long short-term memory for crop prediction

This article introduces a new crop prediction method using hybrid machine learning (ML) and deep learning (DL) models. The proposed model comprises four phases: data preprocessing, feature fusion, feature selection, and prediction. Initially, the dataset is built with the information collected by 250 sensors located at different places in Maharashtra. The constructed dataset has provided sample data for 31 crops, each with four attributes: temperature, humidity, rainfall, and soil potential of hydrogen. After constructing the dataset, preprocessing is the initial step of the proposed framework. Then, feature fusion and selection were performed using the remora-based partial least squares regression method to achieve the best accuracy. Eventually, the most discriminatory features are incorporated into the hybrid ML and DL model known as the extreme learning machine based on the bi-directional long short-term memory for final prediction. The proposed method is implemented in the python platform, and the performance is evaluated in terms of accuracy, precision, recall, F-measure, Kappa, MAE, and log loss. Then, the performance of the proposed method is compared with recent existing methods. As a result, the simulated outcomes proved that the proposed method had achieved better performance than the existing methods.