**1. Introduction**

The advancement in science and machine learning has accounted for a colossal amount of information in the agrarian field subjecting to examination and incorporating procedures such as crop yield forecasting, investigation of plant diseases, enhancement of crops, etc. Machine learning has ascended with enormous processing strategies to conceive new opportunities in multi-disciplinary agrarian innovations. Though machine learning strategies handle immense sums and variations of information, accomplishing a superior performing model is a pivotal plan that needs to be focused. Further, this actualizes the need for scaling the learning model's performance. Feature extraction is a technique for determining a significant subgroup of features utilizing various statistical measures for model construction [1]. It can impact a machine learning model's performance by enlisting a substantial feature subset for boosting the performance and categorizing the variability. The most

prevailing feature selection measures are the filter methods, which are generally faster, and the wrapper methods that are more reliable but computationally expensive.

Together with colossal data advances and improved measure reinforcement, machine learning has risen to determine, assess, and envision intensive information techniques in an agriculture operative environment. Exuberant upgrades in machine learning have tremendous potential results. Many researchers and authorities in present agribusiness are looking at their speculation at an increasingly prevalent scale, helping to accomplish progressively exact and steady forecasts. Precision agriculture is also known as "site-specific agriculture", an approach to deal with farm management utilizing information technology. Precision agriculture assures that the crops and soil receive precisely what they require for optimum health and profitability [2]. Present-day agrarian frameworks can discover significantly more machine learning methods to use enhancements more efficiently and adjust to different natural changes [3–5]. The objective of precision agriculture is to guarantee productivity, manageability, and conservation of the environment [6]. Machine learning in precision agriculture endows a crop management system that assists in yield forecasting in crops, crop disease management, distinguishing the crop weeds, acknowledging crop assortments, forecasting of agricultural climate, and many others. In the machine learning procedure, insignificant features in preparing a dataset will decline the forecasting efficiency [7]. Due to the extensive increase in the data amount, a pre-processing strategy such as feature selection grows into an essential and demanding step when using a machine learning technique [8,9].
