COFFEE DATA SCIENCE

Stability of Grounds Total Dissolved Solids (TDS) Samples in Spent Coffee

Challenging convention

Robert McKeon Aloe
Towards Data Science
3 min readJan 4, 2022

--

A few months ago, I tried an experiment where I put coffee grounds directly onto a refractometer to read Total Dissolved Solids (TDS) as a measure of how many solubles remained in spent coffee. I called this grounds TDS or gTDS to distinguish how the sample is collected. The aim was to get a reading on solubles that might not be easy to extract. A reading did come up, and after some testing, it seemed like a valuable post-shot analysis metric.

Here is a good example where I saw a dark spot in the middle of the puck. I worried it had a lower extraction rate, so I took a few samples. I found out that it actually had a higher extraction rate than the rest as indicated by fewer solubles left in the sample.

All images by author

Some have questioned by I didn’t weigh the water or the coffee when I took such a measurement, and my response was that I didn’t notice a difference in measurement. As a data scientist, that answer can be verified with some data, so collected some data and verified that the amount of water added to such a sample doesn’t matter much. I suspect this is due to the coffee grounds blocking most of the light so as long as they are wet, the reading is the same.

Data

I collected some samples where I weighed the coffee and the water as I took some measurements.

I found some minor changes, but not many, and they certainly didn’t trend well with water added. Typically, if you double the water in a sample, you should expect the TDS reading to drop in half. This was not the case.

Negative values are relative to the calibration of the scale, for which I used the tank water.

From this data, I find that the gTDS reading is pretty stable and doesn’t require weighing the coffee and water as long as the coffee covers the sensor and has water spread throughout.

--

--

I’m in love with my Wife, my Kids, Espresso, Data Science, tomatoes, cooking, engineering, talking, family, Paris, and Italy, not necessarily in that order.