Hi @WeirdBros:

The data source is quite large - around half a million profiles across several harvest seasons. Path Intelligence automatically throws out extreme outliers but there’s also a threshold at which Path Intelligence will not return a suggestion if the sample size is too small to be helpful or if it was created by too few people to be meaningful as guidance. You’ll see this when you choose combinations of certain origins with certain processing methods. If you choose Guatemala with Red Honey as a processing method today, PI will default you back to looking at all honey processed coffees in Guatemala because there is data, but not enough data to be good (or responsible) guidance. So long story short, your “bad” profiles are likely just a drop in the bucket and they are also probably showing up in one of the “less popular” bars on your histogram.

The only thing about Path Intelligence that IS specific to you, your machine and your profile history is in the third pull-down menu where you choose your roaster. When you choose this, you’re telling Path Intelligence to make a suggestion based on those half-a-million profiles, but to normalize the temperature data to better fit YOUR specific coffee roaster. Path Intelligence will look up at what temperature you typically hit first crack on your machine for this country + processing method (because everyone’s thermocouples read slightly different). If you’ve never roasted that combination of country + processing method before, it will look at your typical first crack temp for just the processing method and make the best suggestion. The more your roast, the more it can learn about your machine.

Hope this was helpful. Feel free to call in too, if I’ve misunderstood your question.

Thanks a bunch, Angie! That provided a lot more clarity.