My graph is called a stacked area cohort analysis. Instead of representing my water consumption as fixed units, it shows each type of water consumption as a percentage of my total use that day. I actually made two cohorts, one for my at home water use, and one at school, which are reflected atop one another:
As you can see, my bathing water spiked on day four. This is because I took a bath rather than a shower that day, which took nearly three times as much water. It is this type of insight (all be it on a larger scale) that makes cohort charts useful. For example, in Eric Ries’s new book, The Lean Start-up, it is recommended that cohort analysis be used to track user behavior in a new product.