Estimating Fish Body Weight from Morphological Features Using Some Data Mining Algorithms

The objective was to compare the performance of various data mining algorithms in estimating the body weight of fish. The effects of age, farm type and body morphological characteristics on the body weight of the hybrid silver carp (Hypophthalmichthys spp.) were analyzed using some data mining methods. The body weight of the fish was predicted using different morphological measurements taken during its lifetime and after slaughter. Prediction performances of the Chi-Squared Automatic Interaction Detector (CHAID), Classification and Regression Trees (CART), Artificial Neural Networks (ANN), Random Forest (RF), and Multivariate Adaptive Regression Splines (MARS) algorithms were compared. The prediction capabilities of the fitted models were assessed using model fit statistics. The MARS algorithm was demonstrated to be the most effective model for characterizing body weight. Body length showed the greatest relative importance among all traits measured during the fish's lifetime. When the body length of fish exceeds 145 mm, the body weight is expected to reach 173 g. The MARS algorithm is the most effective model for predicting the body weight of fish, based on the experimental data. It provides an excellent alternative to existing data mining techniques.