Method: Measure the number of airlock bubbles per minute, noting the time and date. Calculate the logarithm of bubble rate, and make a plot of this against time.
This is what is known as a "first order plot". It shows that the rate does not decrease linearly, but that the rate of decrease slows down with time. This is typical of reactions in which the rate depends on how much of the starting material, or reactant, is present. In the case of fermentation, the reactant is sugar. As sugar is consumed by the yeast, the rate of fermentation will obviously decrease. However, the decrease in rate is not linear - the "deceleration" slows down over time. In this type of reaction, when you make a graph of the logarithm of the rate vs. time, you get a straight line. That's what happened here.
Fermentation kinetics are actually a lot more complicated, especially at the beginning of the fermentation process, when yeast populations are growing, and the fermentation rate is accelerating. In this case, fermentation is almost complete and the yeast population is no longer growing. There are lots of yeast cells present, and the rate is only limited by the amount of sugar remaining. The result is "first order kinetics".
One day, I would like to try measuring the fermentation rate from start to finish, but it gets complicated. Measuring airlock bubble rate would be a pain during primary fermentation, when the rate is quite high. The other way you can measure the rate is by looking at alcohol content. The problem with alcohol content is that the measurement is in situ, requiring the insertion of a wine thief and hydrometer into the must. This increases the likelihood of introducing oxygen or microbes into the must.