Very nice article.

I changed "La Niño" to "La Niña," in the introduction.

> If we call this wavelet y(t), all that matters is that it's a bump with wiggles on it, and that its mean value is 0

Perhaps say a bit about why in general the wavelet should have a mean of zero. I see from the wikipedia page [continuous wavelet transform](http://en.wikipedia.org/wiki/Continuous_wavelet_transform) that there are conditions on the wavelet to make the original signal recoverable from the transform, and these imply that the mean is zero. But if we're just searching for patterns in the data, it still looks useful to allow "wavelets" such as a positive pulse.

I changed "La Niño" to "La Niña," in the introduction.

> If we call this wavelet y(t), all that matters is that it's a bump with wiggles on it, and that its mean value is 0

Perhaps say a bit about why in general the wavelet should have a mean of zero. I see from the wikipedia page [continuous wavelet transform](http://en.wikipedia.org/wiki/Continuous_wavelet_transform) that there are conditions on the wavelet to make the original signal recoverable from the transform, and these imply that the mean is zero. But if we're just searching for patterns in the data, it still looks useful to allow "wavelets" such as a positive pulse.