Aw in Specialty Green Coffee Longitudinal Shelf Stability Study

2023. 3. 17. 09:00커피 이야기

반응형


   Finally, we come to the real question that we set out to answer way back when we started measuring water activity.
Does Aw reasonably predict shelf life (storage and shipment stability)?
This question itself is poorly framed.
We mean to ask:

Does Aw measured remotely on a sample pulled from a larger lot of specialty green coffee reasonably predict the post-shipment arrival and storage holding characteristics of that coffee's cup score?

 

   We begin by looking at the same broad water activity categories as we did for the Trans-2-nonenal query.
In this case we substitute average and absolute score change as the target descriptors.
We will look at these both from the perspective of the life of the coffee (PSS > ARR > SPOT > SPOT > etc.), as well as just zeroing in on the change from PSS to ARR.
Average score change (△) can give us a general impression of how a large category performs overall.
Absolute score change takes all change as positive, ignoring the difference between positive and negative (coffees that gained points and coffees that lost points).
This tells us about the degree of change within a category.


   We can see that samples below 0.610 Aw experience less overall score drop, as well as less general volatility.
The coffees above 0.610 lost more points on average and experienced more volatility than the others.
If you remember, 0.610 was roughly one standard deviation above the mean (0.554) for our population.


   The distribution ofour longitudinal set (that is, coffees for which we have multiple measurements taken at time intervals) is very similar to our population estimate.
There is a very slight negative skew that may be the result of increasing selection bias for lower Aw coffees as our work with water activity progressed.


   If we limit the search to look at the score change that occured only between PSSs and their ARRs, we see a similar pattern.
The coffees that began with lower Aw lost fewer points on average and also were less volatile.
One more statistic that we can look at for volatility is standard deviation of score change.
This gives us a view of how diverse the score change values are around our mean score change.


   For the coffees beginning above 0.610 Aw the standard deviation of score change is 2.66 points whi for those beginning below 0.550 Aw it is 1.99 points.
This means (score change is normally distributed) that roughly 68 percent of the samples > 0.610 Aw have score changes between +0.99 and -4.33 points.
Roughly 68 percent of the samples < 0.550 Aw have score changes between +0.94 and -3.04 points.
Both average score drop and score change volatility are greater for coffees that begin with water activity above 0.610.

 

출처 : CAFE IMPORTS

반응형