An Optimized Partitioning Around Medoids Algorithm for Analyzing Agricultural Data Utilizing Deep Neural Network for Feature Selection

In India’s Agriculture, countless decisions are taken by personnel every day. The major challenge is an accurate estimation of crop yield. In the proposed work, the Partition Around Medoids (PAM) algorithm is employed. The algorithm challenges include 1) predicting the k value and 2) comparing cluster quality is time-consuming. This work evaluates agricultural data by 1) selecting optimal parameters and 2) applying optimized PAM (OPAM) to boost crop production. Initially, feature selection is carried out using a deep neural network (DNN). Additionally, massive amounts of crop, soil, and climatic data are mined for processing. Then, analyzing new, non-experimental data is performed which helps in optimizing the production and increases the resilience of agriculture to changes in climatic conditions. Later, the PAM algorithm is used on the dataset. In this case, the user enters a number of clusters (k). The Anarchic Society Optimization (ASO) algorithm selects k automatically. The model is then assessed using performance metrics.