Identify the variables that impact the outcome - like those shown in your third link.  Then aggregate the data into groups for each test set. Usually you want to look at the data as it impacts recovery or grade.

You will create a series of tables for each test set (like pH) and duplicate the data for test sets that overlap.

Once you have done that you can start plotting graphs such as Copper Recovery vs pH or grind.  You should also conduct a regression analysis to determine if the variables identified are statistically significant.  This is not the same as just looking at the R squared value from the graphs.  You may think you see trends but only a regression analysis will tell you if it is real.

Regards