Picking coffee beans could become much easier thanks to a model that predicts whether beans will make good coffee, without having to roast and brew them.
The chemicals found in unroasted green beans vary, and are broken down during the roasting process to release new flavours. Researchers in Parana, Brazil have worked out how to use the chemicals in green beans to predict the taste of the coffee eventually brewed from them. The model they created, published last April in the journal of Food Chemistry, predicted 50% of the difference between coffee samples. “The cup quality of the coffee brews were associated to a set of chemical compounds in the green coffee beans.” Confirmed the researchers.
To make sure the results were relevant to coffee professionals, the researchers used brewing protocols set by the Speciality Coffee Association of America (SCAA) and expert tasters trained by the Brazilian Association of Special Coffee. The researchers used 217 different samples of Brazilian coffee beans when building the model to ensure it was trained with realistic data. Researchers found that beans delivering high quality brews typically had lower levels of caffeine, protein and chlorogenic acids.
Coffee beans are bought in bulk in unroasted form. Geographic origin is the current best quality predictor since climate and soil affect bean growth, however poor weather can ruin taste quality. This model would remove some of the guesswork enabling wholesalers to reassure potential buyers, and buyers to be more confident with purchases.