PREDICTION OF INFECTED SAMPLES IN AGRICULTURAL DATABASE USING DATA MINING TECHNIQUES

Abstract

Data mining in agriculture has come up very recently as a research topic. It is based on using data mining techniques on agricultural data to derive some analysis and patterns out of the data. Many technologies are available these days which help to obtain a lot of information on agriculture and this can be further analysed to get more important information. Data mining in agriculture has a lot of applications like predicting crop yield, predicting the optimum environmental parameters for improving crop yield, predicting the infections in crops etc. By obtaining such useful information we can increase our productivity of crops, we can avoid many problems like infestations in the crops and we can sow the crops according to the optimum conditions. We can try to make the soil more suitable for the crops which are to be grown in them. Hence by using data mining techniques we can extract important and useful information from agricultural data sets which can further be used for important applications and in daily life practical problems. In this project, agricultural data has been analysed which had been provided by the Indian Agricultural Research Institute, New Delhi. The data is regarding various crops imported from various countries over the past 5 years. It focuses on the interceptions (in the form insects, fungi, viruses etc. ) present in the crop samples imported. Data mining techniques have been implemented in R Language over the data to discover interesting patterns and relationships in the data. In particular, clustering algorithms are used to analyse the data and cluster it. From the clusters effort has been made to predict whether a particular crop to be imported from a particular country will be infected with some interception or not and to determine the probability of the samples being infected. On the basis of such result, we can find out whether it will be beneficial for us to import a crop 'A' from a country 'B'.