Calcdifference: Difference between revisions

Latest revision as of 14:16, 30 August 2017

Calculate normalized difference between two datasets.

[diffval, diffvalstr] = calcdifference(x1,x2)

Calculate normalized difference between 2 datasets (scaled to the first dataset input)

diffval = ${\frac {{\text{norm}}\left(x_{1,incl}(:)-x_{2,incl}(:)\right)}{{\text{norm}}\left(x_{1,incl}(:)\right)}}$

where the designation $x_{i,incl}(:)$ specifies that only included data (all modes) are used in the calculation.

@@ Line 1: / Line 1: @@
 ===Purpose===
-Calculate difference between two datasets.
+Calculate normalized difference between two datasets.
 ===Synopsis===
@@ Line 8: / Line 8: @@
 ===Description===
-Calculate aggregate difference in mode 1 of 2 datasets.
-diffval = sqrt(sum((x1i( : )-x2i( : )).^2))./sqrt(sum(x1i( : ).^2));
+Calculate normalized difference between 2 datasets (scaled to the first dataset input)
+diffval = <math>\frac{\text{norm} \left(x_{1, incl}(:)-x_{2, incl}(:)\right)}{\text{norm}\left(x_{1,incl}(:)\right)}</math>
+where the designation  <math>x_{i,incl}(:)</math> specifies that only included data (all modes) are used in the calculation.
 ====Inputs====
-* '''x1''' = previously generated classifier model or pred structure,
+* '''x1''' = dataset to be compared to; difference is scaled relative to this response
-* '''x2''' = 0 or 1. 0 indicates confusion matrix should be based on self-prediction results, 1 indicates it is based on using cross-
+* '''x2''' = dataset to be compared to '''x1'''
 ====Outputs====
-* '''diffval''' = numeric difference value.
+* '''diffval''' = normalized difference between '''x1''' and '''x2'''.
-* '''diffvalstr''' = string of numeric value.
+* '''diffvalstr''' = string representation of '''diffval'''.
 ===See Also===
 [[caltransfer]], [[MCCTTool]]