Multiple regression models for yield prediction in plantain

The regression studies on ten plantain varieties revealed that vegetative growth and fruit characters can efficiently be used for yield prediction in North India. Growth and fruit characters were used separately as predictor variables for developing yield prediction models. Leaf width, plant girth, number of leaves and plant height predicted yield up to 89.29%, while fruit characters, viz., fingers bunch⁻¹, finger diameter, hands bunch⁻¹ and finger length could predict yield up to 99.57% in main crop. On ratoon crop, plant height, plant girth and number of leaves were responsible for prediction of yield up to 78.69%, whereas fingers bunch⁻¹ and hands bunch⁻¹ could predict the yield up to 86.54%.