Coffee Data Science
Does Drink Temperature Correlate to Taste in Espresso?
Examining beverage temperature
I have been exploring beverage temperature and taste for three years now. Initially, I found resting espresso for a few minutes improved taste particularly if the drink got below 50C. After the rise of extract cooling, I collected data and switched to quickly cooling my shots after brewing.
Since then, I have collected more data on drink temperature, and I was curious if temperatures below 50C had an affect on my perception of taste. Of course taste is subjective, and I have worked on being consistent for my own tasting panel. So whatever I find is related to me, but if I find something interesting, it may be an experiment for someone else to try.
Data Metrics
Final score is the average of a scorecard of 7 metrics (Sharp, Rich, Syrup, Sweet, Sour, Bitter, and Aftertaste). These scores were subjective, of course, but they were calibrated to my tastes and helped me improve my shots. There is some variation in the scores. My aim was to be consistent for each metric, but some times the granularity was difficult.
Total Dissolved Solids (TDS) is measured using a refractometer, and this number combined with the output weight of the shot and the input weight of the coffee is used to determine the percentage of coffee extracted into the cup, called Extraction Yield (EY).
Data
I collected data on 535 shots over multiple roasts, multiple days, and changes to shot profile. Most of these were pulled using the Decent Espresso machine or the Kim Express.
I started looking at TDS and drink temperature. The relationship wasn’t linear, and I had noticed on a few shots that higher TDS usually came from more even water flow. So I wonder if measuring water temperature on the basket could give an indication of channeling because channels with faster flow get to the brew water temperature faster.
Then I looked at taste, and it seemed very flat. The scores in the range of 4 to 10 were from robusta coffees before I had better dialed them in.
So I split this by roast, and no easy to spot visual pattern emerged.
I computed the Z-score for taste based on each roast using the mean and standard deviation. This means for each score, I subtract the roast mean and then I divide that by the standard deviation. The value is the number of standard deviations a score is from the roast average (or mean).
This chart also doesn’t show a trend with respect to drink temperature.
I examined drink temperature vs taste, and I didn’t find a relationship. All of these shots were below 50C in temperature, so I wouldn’t extrapolate this trend to hotter beverages.
However, I did see a loose relationship between temperature and TDS. The correlation value between temperature and TDS is -0.39 while it is 0.08 for temperature and taste. For reference, TDS and taste in this data set are correlated at a value of 0.57.
If you like, follow me on Twitter, YouTube, and Instagram where I post videos of espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn. You can also follow me on Medium and Subscribe.
