JosÃÂ© L. RodrÃÂguez-ÃÂlvarez , Rogelio LÃÂ³pez-Herrera, IvÃÂ¡n E. VillalÃÂ³n-Turrubiates, Gerardo Grijalva-ÃÂvila, Arturo Soto-Cabral
This paper proposes two models to estimate individual net weight of the Nile Tilapia (Oreochromis Niloticus), in an experimental intensive production unit located in the temperate zone of the state of Durango, Mexico. As a first approach, it is proposed a model based on the neuro-fuzzy ANFIS system (Adaptative Neuro-Fuzzy Inference System) model, which is used for training phase data size and individual weight obtained from a sampling process during a production period of six months, considering as input variable the precaudal length (centimeters) and as output variable the weight (grams). Different configurations are tested in the antecedent and consequent parameters of the ANFIS network to determine the best fit in the model. The first four months were used to collect data for training, and the remaining two months for validation. As a second approach, a linear regression model using data in the same way (first four months to make the model adjustment and the remaining months to verify the predictive capabilities) is also proposed. The results show that the ANFIS model has greater predictive power, since the error in the forecast inside and outside of the sample is below than the error obtained with the linear regression model.