Boosting Agricultural Productivity Using IoT and Machine Learning-Based Crop Recommendation

Crop production is crucial to the country's economic development by promoting sustainable agriculture. Integrating IoT and machine learning into crop recommendation systems widely contributes towards enhancing product yield, and minimizing farmers' efforts. The correct forecasting of crop recommendation may give a perfect outcome in analyzing the crop features based on its parameters. Objectives: This work aims to improve agricultural productivity using classification models. Our prime objective is to identify the specific parameters like soil productivity, humidity, wind velocity, rainfall, and other climatic conditions that affect the growth of crops. Classification-based prediction provides accurate results based on the optimality of crop characteristics. Materials and Methods: Our hybrid research model shows the efficiency of machine learning and IoT in the domain of crop recommendation systems. We have taken the real-time data of Nitrogen(N), Phosphorus(P), and Potassium(K) values of soil by using the NPK Soil Sensor. The real-time data collection was done at GIET University Agricultural Fields (School of Agriculture), India. The data is further converted into a. csv file for further processing. We have implemented five different types of classification techniques which include Gradient Boosting, Decision Tree, Random Forest, Logistic Regression, and Support Vector Machine(SVM). Result Analysis: The effectiveness of all the different kinds of classification algorithms is assessed using the obtained results. We have estimated the performance metrics (accuracy, precision, F-measure, and recall) of all the classifiers.