Skip to main content
Figure 2 | Veterinary Research

Figure 2

From: Infectious bursal disease virus: predicting viral pathotype using machine learning models focused on early changes in total blood cell counts

Figure 2

Based on data collected at 2 dpi. A Factorial discriminant analysis (FDA) projection of each individual onto the first two components (43 and 18% of the data inertia). Lines link individuals to the centre of gravity of their infecting pathotype group and are associated with their confidence ellipse. Cla gathered the animals infected by the Cla strain, im those infected by im1 or im2 strains, i those infected by i vaccine strain, i + those infected by i + vaccine and Vv those infected by Vv1 or Vv2 strains, B Explanatory variable importance plot considering node purity increase and mean minimal depth [44] for the random forest model associated, C Performances of models in terms of percentages of correctly classified individuals (r:  erythrocytes blood concentration, thr: thrombocytes blood concentration, m : monocytes blood concentration, g:  granulocytes blood concentrations, b: B lymphocytes blood concentration, t: T lymphocytes blood concentration, n:  bursal viral load, u: uric acid blood concentration, CS: clinical score), d: dimension.

Back to article page