Usually it’s for 2 things that change together. Like temperature & ice-cream sales.
I use graphs for insulin against change in BGL
VS
CHANGE in BGL is important.
I want to know what amount of insulin (for a specific meal) will give me a ZERO change in BGL (i.e. where the diagonal line cuts through the horizontal axis).
I note what I see as dots on the graph. Eventually I get a better and better ‘line’ connecting the dots.
Sometimes, a dot might land in a strange spot. I still note it, but my main aim is to get a line that fits the most dots (I won’t worry if it can’t fit all of them).
Lastly, I find the graph good for correcting hypers too.