I wasn't familiar with Kalman, and I tried to understand it by reading the primers on this website and it wasn't clicking for me. Last night I was thinking about the sample graphs, and it occurred to me that it just looked like an Exponential Moving Average, which is a much simpler concept to grasp.
I just searched Google for a comparison and found that EMA is as good as Kalman for "random walk plus noise" on a Stats Stack Exchange, and that a 2003 paper from Brown University (Joseph J. LaViola) showed that a Double Exponential Smoothing algorithm is of equal quality to Kalman (and extended Kalman) but 135 times faster, and a simpler approach.
I find Double Exponential Smoothing to be much easier to understand than Kalman, and assuming the LaViola paper is correct, I'm not going to put additional effort into understanding Kalman.
I just searched Google for a comparison and found that EMA is as good as Kalman for "random walk plus noise" on a Stats Stack Exchange, and that a 2003 paper from Brown University (Joseph J. LaViola) showed that a Double Exponential Smoothing algorithm is of equal quality to Kalman (and extended Kalman) but 135 times faster, and a simpler approach.
I find Double Exponential Smoothing to be much easier to understand than Kalman, and assuming the LaViola paper is correct, I'm not going to put additional effort into understanding Kalman.