**Voicemail transcript** > **From:** Luciana Moretti > **Date:** Thursday, 9:47 AM > **Duration:** 2 minutes 14 seconds > > Hi -- this is Luciana Moretti, from Bodega Moretti in Mendoza. I'm calling about the data analysis work. > > So, here's my situation. Every harvest I taste through -- what is it, three hundred? Maybe three hundred and fifty barrel samples. And I have to decide which ones become Reserve and which go into the standard line. The Reserve sells for four times the price, so getting it right matters. > > But I know I'm inconsistent. Some years I'm generous, some years I'm strict. My head winemaker tells me the same thing. And honestly, I'm getting too busy to taste everything as carefully as I should. The new export contracts mean more production, which means more barrels, which means more tasting. > > What I have is five years of production data. Fermentation temperatures, which vineyard plot -- we have three at different altitudes -- rainfall during growing season, soil analysis, how long each barrel aged, and my quality scores from the blind tasting panels. Maybe three thousand barrel samples total. > > What I want is a model that can predict which barrels are likely to score high enough for Reserve based on the production data. Not to replace my tasting -- I will always taste -- but so I can focus my time on the borderline cases instead of tasting through barrels that are obviously standard. > > I also want to understand what production factors actually drive quality. I have my instincts -- altitude matters, I'm sure of that -- but I want to see the data. > > And it needs to be something I can explain to my export partners. They always ask how I select Reserve. "I taste everything" is true but it's not very... convincing as a method. Something more systematic would help. > > Send me a message when you're ready to start. I'll share the data files. Gracias.