Well, that is a fair point.

I mean long-term; over decades. It's my impression that Coke is everywhere, and nowhere left to expand. The market is saturated.

If people realized the harm sugary drinks do to them, they wouldn't buy any. But if they do, there are plenty with almost the same taste as Coke, but at least 2x cheaper.

I think the choice presented is interesting. But I don't think the chart is made beautiful by the data, which is what this sub is about.

Dividends are when a company pays out profits when it has nothing better to invest them in.

Coca Cola has no growth potential, all they do now is cannibalize what's left of the branded plastic sugar water industry.

> the value quickly gets NaN

Sounds like numerical issues. That could be caused by a too high learning rate.

What does your training error look like across iterations? If it jumps all over the place (increasing a lot maybe), then it's too high, as the step sizes overshoot their targets repeatedly.

Those scales typically rely on your data (your age and height and whether you're an athlete) more than the impedance measurement.

In the case she does not like the smut, she can get over it.

If she can't get over it, maybe she's not worth keeping as a friend.

If she tells you she enjoyed the book, you will find a deeper friendship.