PRATIBHA BHARTI*, SARITA RANI, AJAY SHARMA, PUMMY KUMARI
Department of Mathematics and Statistics, CCS Haryana Agricultural University, Hisar-125 004 (Haryana), India
Deparment of G&PB (Forage section), CCS Haryana Agricultural University, Hisar-125 004 (Haryana), India
*(e-mail: pratibha595@gmail.com)
(Received : 07 March 2025; Accepted : 30 March 2025)
SUMMARY
The present study was carried out for classifying and predicting classes for yield and
protein content of multicut forage sorghum genotypes using discriminant analysis, based on
performance measures derived from a confusion matrix. Secondary data of 117 genotypes of multicut
forage sorghum, along with two checks measured for 15 traits, was used in this study. The genotypes
were grouped into two categories, G1 (low) and G2 (high), under two grouping schemes (GS I and GS
II) across two datasets: 1st cut and 2nd cut. Classification and prediction results were obtained for
both training and testing datasets. A confusion matrix was generated from the testing data to classify
and predict classes based on fodder yield and protein content. The highest accuracy percentage
(85.7%) was achieved in grouping scheme GS I for green fodder yield (GFY) in the testing dataset of
the 1st cut, demonstrating the effectiveness of discriminant analysis in accurate classification and
prediction.
Key words: Discriminant analysis, performance measures, green fodder yield, accuracy, confusion matrix