Penerapan Model Regresi Bertatar dalam Penentuan Hasil Tanaman Sorgum
Sorghum is one of the prospective food crops to be developed in Indonesia. High yielding sorghum variety development is carried out through a series of breeding processes, including agronomic character selection among sorghum plants. The research's objective was to examine the stepwise regression model in selecting agronomic parameters that affect the yield of sorghum grain yield. Evaluation and modeling of the relationship between agronomic parameters and sorghum yield was carried out with two varieties of sorghum, namely Super 1 and Super 2, by using an experimental engineering approach. The estimation model based on information on the phenotypic character of sorghum was implemented using stepwise regression analysis. The results indicated that agronomic parameters affect the yield of sorghum was estimated in three stages, and the final yield equation is Y = -3.413 + 0.033 x2 + 0.561 x3 + 0.006 x5. The variables included in the model are x3 (panicle width), x5 (plant height at harvest), and x2 (panicle length). Correlation and determination coefficient values up to this stage were 0.82 and 0.68, respectively. This indicates that sorghum production can be optimized by the three significant variables: viz, panicle width, plant height at harvest, and panicle length. These results can then be used as a basis for selecting varieties to obtain high yield potential varieties.