ABSTRACT

Our nation India is an agrarian country and its economy based upon agricultural crops production. The share of agriculture in GDP increased to 19.9 percent in 2020-21 and almost fifty percent of total manpower utilized their efforts in this sector. The uncertain whether, climate changes and water storage with traditional farming trends and improper irrigation facilities are directly affect the crop productivity. All such parameters make the environment of uncertainty regarding crops production. On the other hand, accurate and timely predictions of crop production are backbone of the policy maker regarding import-export, demand-supply, marketing, pricing and distributions to balance the socio-economic frame.

The uncertainty and its prediction tend to be complex phenomena. The primary resources of uncertainty are randomness and fuzziness. The randomness deals with general uncertainties while fuzzy logic suitable for the complex phenomena. The statistical regression methodology is used traditionally for such complex predictions. In series of smart notations, the smart forming is necessity of day. Hence there is a requirement to develop qualitative and statistically sound prediction of crop yields with machine learning handling large amount of data.

The present works focus on investigation of various used machine learning algorithms for their suitability in crop yields perdition and finally proposed an approach based on linear regression and fuzzy logic in big data computing paradigm for accurate and timely predictions of crop production.

Keywords: Uncertainty, Computational Intelligence, Linear Regression, Fuzzy Logic, Time Series Analysis

| DOI: 10.17148/IARJSET.2021.8855