Benchmarking Machine Learning Models for Predicting the Average Weight of Oncorhynchus Mykiss in High Andean Fish Farming (#2262)

Authors

Uceda Martos, Patricia Janet

Sanchez Quiroz, Samir Joseph

Chavez Huaman, Santos Roel

Ruiz Regalado, Reyles Aly

Torrel Villanueva, Manuel Eduardo

Abstract

The present research compares Machine Learning models to predict growth (average final weight) and support profitability decisions in a high Andean fish farm. Linear and Random Forest Regression were evaluated using a set of 500 weekly records, integrating historical data provided by the fish farm and synthetic data generated within validated physicochemical ranges. The results concluded with a high performance in the growth prediction (R²>0.98) and adequate performance in profitability (R²≈0.87). In addition, the profitability classifier achieved 0.93 accuracy, suggesting operational utility as an early warning.