A Binary Logistic Regression Model for Prediction of Feed Conversion Ratio of Clarias gariepinus From Feed Composition Data

Aquaculture in developing countries faces a lot of challenges that are barely being addressed. With feed taking nearly 70% the total production cost, it becomes imperative to develop means optimizing how research is conducted into development. Feed conversion ratio as measure quality can be used quantify retrospect appropriateness fed livestock, particularly, Clarias gariepinus. From study, binary logistic regression simple terms, determine if prospective will perform below or above acceptable level 1.5, based on its composition and proximate analysis values. Data from similar experiments normalized split train testing data fit model, three numerical optimizers were including liblinear, Newton-CG, SAG accuracy models compared using confusion matrix, Jaccard similarity score. An value 0.8 was observed model regardless optimizer, this indicates predicting either high low FCR for types. The probability prediction showed disparity among liblinear SAG/Newton-CG solvers. Liblinear solver close probabilities values 1 0. While made by all solvers, possible affinity error when used. This also indicated with logloss 0.65 0.51 both Newton-CG