Mining a tea insect pests database

Data mining techniques are being applied successfully in wide varieties of databases in order to extract useful information. This paper applies data mining techniques on a new tea insect pests database created on the basis of data available from different tea gardens of North Bengal districts of India. We describe different issues related to the development of a good data mining model in the present context. We propose a novel multiple imputation - reduced dimension - clustering approach. A bootstrap-based EMB algorithm performing multiple imputation for missing values; EM- clustering technique; Id3 and C4.5 for tree based classifications have been deployed in the study. The performance of the model is found satisfactory.