RESUMEN
The objective of this research is to apply exploratory analysis and modeling associated with abiotic factors, physiological and behavioral variables of swine in the semi-arid region. The experimental design used was completely randomized, in a 3 × 3 factorial scheme, randomly distributed in nine pens, with three animals. The behavior of the animals was recorded using images and analyzed within 10-min interval. The data analysis used was multivariate, using the clustering method (tree diagram) and principal component analysis (PCA), in order to establish the main predictors of swine ingestive behavior, using multiple linear regression models. The PCA showed satisfactory results, in which the lowest eigenvalue observed was 2.82 and the accumulated variance for the treatments ranged from 69.70 to 94% for the first two principal components. Through exploratory data analysis, it was possible to identify the relationship between biotic and abiotic factors with the ingestive behavior of pigs in the finishing phase. Based on the results of the multivariate analysis, the most promising predictor variables for estimating the regression models were determined. Adiabatic evaporative cooling associated with 18 h of light was the combination of factors with the best results, presenting models for eating and drinking behavior, i.e. a complete ingestive characterization.
