That eamonn guys sounds smart.

If you do not normalize, small differecnes in the mean and/or STD rapidly swamp any shape similarity.

Here is one way to think about it.

If you do not normalize, and the means or STDs are different, then.......

EuclideanDIST(A,B) is approximately equal to DTWdist(A,B), which is approximately equal to [abs(mean(A)-mean(B)) times a constant]

If you can model the similairy with just the mean, they why do you need to even look at the shape?

BTW, you have a cool problem/dataset

That eamonn guy sounds smart.

If you do not normalize, small differences in the mean and/or STD rapidly swamp any shape similarity.

Here is one way to think about it.

If you do not normalize, and the means or STDs are different, then.......

EuclideanDIST(A,B) is approximately equal to DTWdist(A,B), which is approximately equal to [abs(mean(A)-mean(B)) times a constant].

If you can model the similarity with just the mean, then why do you need to even look at the shape?

BTW, you have a cool problem/dataset.