Soul of twit, more like.

MAPE can be a problem also if you have a problem where rare excursions are what you want to predict and the cost of missing an event is much higher than predicting a non-event. A model that just predicts no change would have very low MAPE because most of the time nothing happens. When the event happens, however, the error of predicting status quo ante is much worse than small baseline errors.

My reading of this situation is that MAPE would do the opposite. Means are skewed towards outliers.

"just a tooling issue/a paradigm mismatch" is how many people pronounce "bad UX".

Helion is legitimate and they have a very clever approach, but it definitely still isn't a sure bet that they will succeed.

Check out the ski train from Denver to Winter Park.

LA-SF is nearly 500 MILES. That's close to 800km and driving through LA makes it feel like 1000km. If the sleeper would extend on to San Diego, this would be a sweet item for me.

Google Maps gives LA-SF straight line distance as 347 miles and driving instructions (via I-5) as 381 miles.

AND if you do the zephyr during the full moon, you will see the salt flats in the moonlight. That's definitely a reason to stay up all night.

Check out how they work.

One kind emits light through a folded optic path through the test gas and detects the difference in the absorption between wavelengths of light are used. The difference in absorption is minute so the optical path needs to be as long as possible which makes it hard to make the detector small.

The other kind detects the sound emitted when a precise frequency of IR is transmitted through a gas. If the frequency of the light is just right, it will be absorbed by the CO2 causing it to heat up ever so slightly which causes a tiny bit of sound.

Both of these are incredibly ticklish devices to design and build in mass.

That you can get these for as low as about $20 in quantity is actually mind bogglingly cheap.

Interesting. I have been running three of their devices of different generations in my house for several years now with no hangs.

No idea why. I tried to change the power supply, and even slightly overvolt the power (to 6V from nominal 5V).

Tell me how you will write a unit test for determining the percentage of your utilities that is deductible on Schedule C.

Or a unit test that determines whether the discussion during a meal with a customer was "substantially about business matters".

You're conflating determination of factual and legal questions (out of scope) with modeling the decision tree (in scope and useful).

The function you ask about would be "getDeductiblePercentage()," and the unit tests would return various hard coded numbers. Actually determining that value for a real taxpayer is still hard.

Being able to show how information flows through the US tax code would be useful, even if it doesn't solve all the problems that arise from its intricacy.

What percentage of US citizens expense business meals? This is all about automating the simple cases.

Anything more complex is an input to the system, the system can still be tested.