XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 939
Figure 10 presents the decision tree built in Figure 9
with model statistics. The major split is around 0.19% Cu
in the feed: higher feed Cu leads to higher Cu in the Pb
concentrate, lower leads to lower Cu in the Pb concentrate.
These splits overall agree with the contours from the neural
network model.
Chemical Reagent Addition—Introducing RSM with
Factor Interactions
RSM is particularly useful for understanding interactions
among flotation variables, like interactions between reagent
(CuSO4) and ore (Zn in mill feed) and their impacts on
Figure 9. Plot of decision tree predictions and actuals for the same data set as used in Figures 6 and 7
Figure 10. Decision tree of the same data from Figures 6 and 7
Figure 10 presents the decision tree built in Figure 9
with model statistics. The major split is around 0.19% Cu
in the feed: higher feed Cu leads to higher Cu in the Pb
concentrate, lower leads to lower Cu in the Pb concentrate.
These splits overall agree with the contours from the neural
network model.
Chemical Reagent Addition—Introducing RSM with
Factor Interactions
RSM is particularly useful for understanding interactions
among flotation variables, like interactions between reagent
(CuSO4) and ore (Zn in mill feed) and their impacts on
Figure 9. Plot of decision tree predictions and actuals for the same data set as used in Figures 6 and 7
Figure 10. Decision tree of the same data from Figures 6 and 7