Consider the following outputs based on fitting a weighted arithmetic mean (WAM) and a weighted power mean with p = -2 (PM)

WAM

RMSE 0.1513 Average L1 error 0.1307 Pearson correlation 0.8372 Spearman correlation 0.7541 i w_i 1 0.1266 2 0.3733 3 0.2085 4 0.2916

PM

RMSE 0.1199 Average L1 error 0.1249 Pearson correlation 0.8356 Spearman correlation 0.7174 i w_i 1 0.0769 2 0.3847 3 0.2308 4 0.3076

Which of the following could be stated about the data that was used for fitting, if we prefer minimising the gaps between the fitted value and true value? [Multiple answers could be selected]

Question 6 options:

	Better fitting is achieved for functions where the output variable in the data tends toward the lower inputs.
	Better fitting is achieved for functions where the output variable in the data does not tend toward the lower inputs.
	The most influential/useful variable in predicting the output is x2
	The highest input is allocated the most weight.