And the human crash rate is in all times of day, and all forms of weather. The rate for daytime, clear weather driving is almost certainly even better than 1 per 200k vehicle-miles.
He was referring to ACSOI or whatever it was called, that tried to show profitability by subtracting out marketing costs, in a marketing-heavy organization.
You're referring to Groupon's decision to treat cash they collect on behalf of merchants (which should be treated as financing cash flows) as operational cash flows, thus making their cash flow look reasonable when it's actually horrid.
Counterclaims are the standard response to an initial filing. This is exactly what everyone, everywhere who has even a passing familiarity with litigation in the United States expected.
I used the wilson score method to generate star ratings off of individual user up/down reviews on a commerce site thirteen years ago.
You can do something very similar with plaintext reviews by doing sentiment analysis to mark them as a yay, a nay, or a no opinion, and generate a similar score.
There are other, more complicated, ways of doing things, but this is fine for somebody who only has a modest knowledge of stat.
The trouble is that a lot of NDAs include onerous clauses. I was recently handed an NDA that would've barred me from working in the potential client's entire vertical if I signed it. (yep, a hidden exclusivity clause, in a fucking NDA)
I don't mind signing things, but they need to give them to me well in advance so I can review it with counsel and make sure it's something I'm comfortable signing.
A test needn't be purposefully bigoted to be effectively bigoted.
If I were to sit down and try to create a test of intelligence right now, I'd almost certainly include dozens of biases based on my own abilities, perceptions and experiences. Many of these biases would translate into higher scores for people like me, and lower scores for those whose lives differed.
This would create a situation where, either by malice of negligence, the process would be bigoted.
----
Sperazzo published a paper in 1959 documenting racial differences between Negros and Whites in these tests, and also noting as a previous given that "the performance test is sensitive to socioeconomic background".
As such, it seems like it was known, at the time, that it wasn't so much a test of intelligence but a test of race and socioeconomic status.
Something I've seen clearly is that if you haven't talked quietly with the founders of a startup, you might not know what they're doing.
I'm building a company right now that intends to have three major Acts in it's lifetime, the first of which is unimpressive but designed to get us customers and base knowledge for use while executing Act II, which we then intend to apply to additional verticals as part of Act III.
But Act I will likely look like something that two guys could put together in their garage in a year. And that's exactly what it is.
All the biggest open-source contributors I know made many of their contributions while collecting a paycheck. That doesn't make the work less valuable.
I kicked in a few bucks to the project because just earlier today I wanted to run a small sintra project and had to be walked through a complicated toolchain installation that wasted time and caused frustration.
Here's to hoping he can make it legitimately simple again!
Unfortunately, it's limited to the website, and doesn't carry over to Roku/AppleTV/etc.