--- a
+++ b/Semantic Features/RMSE.m
@@ -0,0 +1,10 @@
+function [ RMSE ] = RMSE( X, Y )
+%RMSE Root Mean Squared Error
+%   RMSE is a standard method, wiki it. Its more 'relatable' to real values
+%   than MSE so using this method to get an idea how bad our predictions
+%   are off. Only makes sense for continous numerical labels, not classes.
+
+RMSE = sqrt(mean((X - Y).^2));
+
+end
+