Gujarat Journal of Extension Education (ISSN 2322-0678)

Title: ATTITUDE OF PG SCHOLARS OF AGRICULTURAL EXTENSION TOWARDS APPLICATION OF MOBILE TECHNOLOGY USING ARTIFICIAL INTELLIGENCE TECHNIQUE

Authors: R. S. Parmar, N. M. Veged and Vishal Mehra

Publisher: Society of Extension Education Gujarat

Keywords: agricultural extension, mobile technology, artificial intelligence, machine learning

Volume: 33

Issue: 1

Year: June 2022

DOI: https://doi.org/10.56572/gjoee.2022.33.1.0028

Abstract: Application of mobile technology is playing a vital role for the enhancement of understanding of agricultural market conditions and farmers business towards agricultural. In present days, data are growing rapidly in massive amount in every domain. One such domain of interest for researchers is application of mobile technology in agricultural extension field. To find interesting hidden patterns from the experimental datasets, artificial intelligence technique is used to build accurate algorithms for classification and prediction. In this paper classification and predictive models for attitude towards application of mobile technology in agricultural extension of the postgraduate scholars are built using machine learning classification algorithms viz; functions based multilayer perceptron and support vector machines; lazy based k-nearest neighbors and KStar; tree based J48 and random forest with respect to their accuracy of correctly classified instances, incorrectly classified instances and receiver operating characteristic (ROC) area. Experimental results explain that random forest algorithm is better than other fitted algorithms with 81 % predictability, followed by k-nearest neighbors with 78% predictability and Support vector machines algorithm has the lowest predictability with 69 %. The study also suggested that predictor variables namely extracurricular activities, information collection interpersonal communication and professional zeal have significant influence on attitude towards application of mobile technology in agricultural extension of the postgraduate scholars. Based on all the benchmarks used to measure the machine learning algorithms fitted in this study, it was discovered that Random Forest algorithm performance is the most appropriate in terms of predictability based on experimental dataset. Therefore focus was to design a predictive system on the most suitable algorithm which is random forest in this domain.

Full Text

Photo Gallery