For my meals out, I prefer to eat at restaurants which publish nutrition information, but I'm under no illusions that these values are anything more than rough guidance. The core of the problem is in employee training. I find that even restaurants which make an effort to publish nutrition info don't bother to adequately train employees to serve the expected amounts of food.
Your cheese example is a good one. Based solely on anecdotal evidence, I'd wager most people who order cheese at Chipotle end up being served 1.5-3x the intended serving size due to generous and unmeasured portions. I love cheese as much as the next guy, but an extra few hundred unexpected calories adds up very quickly. This also renders any analysis of nutritional value misleading if not useless.
There are a ton of interesting datasets at InfoChimps, but many are almost completely unstructured. They require a lot of preprocessing and natural language work to ready them for analysis. It's usually worth it, though, and the ETL coding has really helped me grow with Python.
For my meals out, I prefer to eat at restaurants which publish nutrition information, but I'm under no illusions that these values are anything more than rough guidance. The core of the problem is in employee training. I find that even restaurants which make an effort to publish nutrition info don't bother to adequately train employees to serve the expected amounts of food.
Your cheese example is a good one. Based solely on anecdotal evidence, I'd wager most people who order cheese at Chipotle end up being served 1.5-3x the intended serving size due to generous and unmeasured portions. I love cheese as much as the next guy, but an extra few hundred unexpected calories adds up very quickly. This also renders any analysis of nutritional value misleading if not useless.