Using stats instead of sensors

Statistics often get a bad rap in math circles for being inconclusive at best and downright misleading at worst. While these sentiments may sound true for things like political polls, they hide the fact that statistical methods can be put to good use in engineering systems with great results. [Mark Smith]for example, has worked on an espresso machine that can make the perfect shot of coffee, and turned to one of the tools in the statistics toolbox to solve a problem rather than add another sensor to his complex coffee maker.

To make espresso, steam is generated which is then forced through finely ground coffee. [Mark] discovered that his espresso machine often poured too much or too little coffee, and to improve his machine’s accuracy in this area, he turned to the linear regression parameter R2, also known as the coefficient of determination. Using a machine learning algorithm tuned to this value and assessing predictable variation in a data set makes it easier for a computer to tell when the coffee starts pouring out of the portafilter and into the espresso cup based on the pressure and water flow in the cup. machine itself instead of using another input, such as the weight of the cup.

We have seen in the past how serious [Mark] takes his coffee maker, and this is another step in a series of improvements he’s made to his equipment. In this iteration, he also produced a simulation in JupyterLab to help him better model the system and make even more accurate predictions. It’s a bit more effort than adding sensors, but since his espresso machine already contained quite a bit of computing power, it’s not too big a jump for him.

This post Using stats instead of sensors

was original published at “”