Soil Classification & Prediction of Crop Status with Supervised Learning Algorithm: Random Forest

Abstract

Crop Management System (CMS) as developed in an Ionic framework with a Real-Time Firebase database for loop backing and decision support. The main two features were; Soil classification where the soil classified based on temperature, humidity, and soil properties such as soil moisture, soil nutrients, and soil PH level using Random Forest Algorithm. By Bootstrap method using Random Forest, samples from the dataset were selected & then classification trees was generated. The other feature was crop precision where the condition of the crop was and examined using temperature, humidity, soil moisture, soil PH levels, and soil nutrients (N, P, K). IoT device was used to fetch data from the field and then compare with already stored ideal values, suitable for optimal yield, in CMS database then process using the application to suggest the crop for cultivation and to optimize the usage of water and fertilizers. Currently, we classify the soil using Random Forest Algorithm & suggest the suitable crop for the classified type of soil & also measure the soil moisture and soil nutrients of agricultural field Acre based on the reading results we are suggesting the crop to is cultivated and pre-requisite which would be needed in future. The proposed method gives an accuracy of 96.5% as compared to existing methods of Artificial Neural Networks and Support Vector Machines