Statistical analysis alternatives to validate hematic parameters in dairy under tropical conditions

The hematic parameters of cows selected for milk production in grazing under tropic conditions were analyzed based on six experimental studies. The major objective was to determine statistical analysis alternatives that could be used in the study of physiological, environmental or genetic variations. The parameters analyzed were hematocrit, hemoglobin concentrations and counts of polymorphonuclear cells (i.e., eosinophils, basophils, neutrophils and lymphocytes). Initially 972 data from adult cows with productions considered medium-high in the context of local dairy production were analyzed. After the information decanting process, 415 records were used. These data corresponded to a sampling period from 30 days before calving to day 105 postpartum, with determinations made fortnightly.  Blood collection was made every two weeks and each moment was defined as a period. Different analytical techniques were applied in searching to extract the best information from the data and that could be reused in the future. The best type of analysis corresponded to the Random Forest (RF) technique, clustering, and analysis of variance.The effect of an indicator of energy metabolism (non-esterified fatty acids - NEFA), an indicator of protein metabolism (total protein - TP) of 10 collection periods, three breed groups and three types of nutritional management given to the animals on the blood variables were analyzed. Multivariate correlation and machine learning methods were used to extract information from the data. Results indicated that hematological behavior change throughout close up dry cow, transition period and first phase of lactation. No high direct or inverse correlation were found among variables. Hematocrit and hemoglobin levels can be estimated from variables associated with metabolic indicators and blood components. Further studies are required to elucidate the hematic behavior of dairy cows during the transition period.