I'm pretty familiar with the USDA SR dataset, and here are my thoughts (examining data with one of my experiments diarytail.com)
1) People in the enterprise space will be willing to deal with the weird nuances of the USDA-SR data. For example writing 'beef cooked' vs' 'beef raw'.
2) In the general consumer space, there is a huge problem in getting the general public to find ingredients they use. I found lots of people typing in brand name ingredients not matching what is in the USDA data set.
To me, the problem is partially solved with My Fitness Pal's huge dataset of user generated data. However the flip side is it is missing most of the micros as they are entered via a nutrition label.
Another solution might be to reference nutritionix.com 's dataset (I think the joyapp.com uses it also)
Unfortunately nutritionix has only basic nutritional data as well, the sort of thing you find on a product label. Not bad but not great. Have a look at their sample data. Plus it's around 10k/year or 50k/license.
I'm thinking home cooks would get a lot of value from the tool, and reasonably good data even if they use raw instead of cooked (as an example).
1) People in the enterprise space will be willing to deal with the weird nuances of the USDA-SR data. For example writing 'beef cooked' vs' 'beef raw'.
2) In the general consumer space, there is a huge problem in getting the general public to find ingredients they use. I found lots of people typing in brand name ingredients not matching what is in the USDA data set.
To me, the problem is partially solved with My Fitness Pal's huge dataset of user generated data. However the flip side is it is missing most of the micros as they are entered via a nutrition label.
Another solution might be to reference nutritionix.com 's dataset (I think the joyapp.com uses it also)