Why machine learning hates vegetables | Emily Riederer

[utterances-bot]

Why machine learning hates vegetables | Emily Riederer

A personal encounter with 'intelligent' data products gone wrong

https://emilyriederer.netlify.app/post/ml-vegetables/

[hlostam]

Thanks Emily for a nice article. At first I thought, that you might have forgotten to report one single day in the app. In that case a we would see a pattern such as "when you eat your calories are higher" or "when you record your data your calories are higher". But then you exclude this option saying you did it meticulously. I am betting on association rules as well, ordered in a way that prioritise these highly occuring items.

[ChrisDStats]

I know someone who ate to her glucose meter and put her diabetes in remission. Basically she cut out carbs. If you eat to your meter you can figure out carbs don't help diabetes without knowing anything about the role of carbs like bread and cereal and pasta in raising blood sugar and insulin. So yes timely feedback that allows you to link causes (what you eat) to outcome (blood sugar) can be very useful in managing conditions. AI can be supplied by the human being getting the feedback. :-)