ICAR-IASRI Newsletter, July-September 2014

Abstract

An approach based on RF methodology was proposed (combined Random Forest) for the prediction of disease risk from imbalance case-control data. The proposed approach was compared with the existing methods meant for imbalance data, namely, Balanced Random Forest (BRF) and Weighted Random Forest (WRF) based on performance metrics, viz., sensitivity, specificity, classification accuracy, and precision. The proposed approach was illustrated using a case-control data set of Rheumatoid arthritis and found to perform better in terms of prediction accuracy over the existing BRF and WRF. Some methods of constructing designs for breeding trials involving complete/ partial three-way crosses have been developed. Three-way crosses are intermediate between single and double cross hybrids with respect to uniformity, yield, stability and the relative simplicity of selecting and testing and hence are increasingly gaining popularity among breeders. Series of designs involving complete/ partial three-way crosses for test vs. control line comparisons have also been obtained. An empirical best predictor for small area means has been developed which addresses the problem of small sample sizes as well as dissimilarities between the small areas. An approximately unbiased estimator for mean squared error of the empirical best predictor of small area means is also proposed.