Data Analytics for Crop Recommendation towards Sustainable Agriculture Development

Abstract

Traditionally, farmers used the natural signs of fertile soil, rain and temperatures and sunlight to direct the planting and harvesting. Nevertheless, traditional agronomic systems of weather prediction, seasonal patterns and soil fertility have been seriously compromised since the general pollution that commenced in the 1950s. The world today thus requires the solutions of the present day. Class imbalance, i.e., some categories of crops or soils are underrepresented, is a problem that can be found in the dataset of agriculture. Such imbalance can be reduced with the help of methods Synthetic Minority Over-sampling Technique [SMOTE], which helps to increase the number of minority classes. Results indicate that the quality of classification may be advanced in a great way due to the methods of data mining and Machine Learning [ML], especially when SMOTE is used. Modern agricultural science and technology have supplied the farmers with effective mechanisms to overcome the issue of low productivity through soil testing, crop monitoring and making prudent decisions. Finally, this study will aim at helping farmers to overcome the agricultural problem and embrace the best practices to help both the small scale farmers and the farming managers to choose the crops with the highest yield.