Investigating the "Using all results from Fitting" approach
Maybe using all the data from the Fitting results to compute the source localization of templates is a bit counter-productive at some points? Instead, using as much relevant data as possible could make more sense?
For example, using only data with correlation above 80%. Or using a weighted mean, investigating various weighting formulas, like Correlation, Kendall Tau, Rank etc...
Investigating the "Using all results from Fitting" approach
Maybe using all the data from the Fitting results to compute the source localization of templates is a bit counter-productive at some points? Instead, using as much relevant data as possible could make more sense?
For example, using only data with correlation above 80%. Or using a weighted mean, investigating various weighting formulas, like Correlation, Kendall Tau, Rank etc...