The p Value and the Base Rate Fallacy

Alex Reinhard, on his Statistics Done Wrong site:

You’ve already seen that p values are hard to interpret. Getting a statistically insignificant result doesn’t mean there’s no difference. What about getting a significant result?

As an aside it’s a wonderful site. Read it in full if you have the time.


“Swift will be open source later this year”

Awesome move by Apple.


The Online Privacy Lie Is Unraveling

Natasha Lomas, writing for TechCrunch:

Key findings on American consumers include that —

  • 91% disagree (77% of them strongly) that “If companies give me a discount, it is a fair exchange for them to collect information about me without my knowing”
  • 71% disagree (53% of them strongly) that “It’s fair for an online or physical store to monitor what I’m doing online when I’m there, in exchange for letting me use the store’s wireless internet, or Wi-Fi, without charge.”
  • 55% disagree (38% of them strongly) that “It’s okay if a store where I shop uses information it has about me to create a picture of me that improves the services they provide for me.”

Interesting figures. The tide may end up turning after all.


How Browsers Work: Behind the Scenes of Modern Web Browsers

Tali Garsiel and Paul Irish:

Tali published her research on her site, but we knew it deserved a larger audience, so we’ve cleaned it up and republished it here.


Organizational Debt Is Like Technical Debt — But Worse

Steve Blank:

Just when things should be going great, organizational debt can turn a growing company into a chaotic nightmare.

Growing companies need to understand how to recognize and “refactor” organizational debt.


Exercise Alone Won’t Make You Lose Weight

Aseem Malhotra, writing in the Washington Post:

The idea that our obesity epidemic is caused by sedentary lifestyles has spread widely over the past few decades, spurring a multibillion-dollar industry that pitches gadgets and gimmicks promising to walk, run and kickbox you to a slim figure. But those pitches are based on a myth. Physical activity has a multitude of health benefits — it reduces the risk of heart disease, Type 2 diabetes, high blood pressure and possibly even cancer — but weight loss is not one of them.

Spoiler: eat less refined carbs. In particular sugar:

And no, artificial sweeteners are not a good substitute.


I Was an Undercover Uber Driver

Emily Guendelsberger, for the Philadelphia City Paper:

I talked to lots of drivers. But few kept a meticulous enough log of hours worked, miles driven and expenses paid that I felt comfortable using their data alone. Many drivers worried about getting in trouble, too — Uber can “deactivate” a driver for any reason. I needed someone on the record, someone whose data I knew I could trust.

So, in January, I applied to be an UberX driver myself.

Sobering story. Especially this bit:

On top of that, there’s periods where it’s unclear if anyone’s insuring the driver.

“Technically, Uber’s insurance covers you from the time you pick your passenger up until the time you drop them off,” says Jason Fidishun of Yardley-based Fidishun Insurance, an independent agency representing Progressive, which he says offers the only hybrid personal-commercial policy available in Pennsylvania developed with Uber and Lyft drivers in mind. The driver’s personal insurance, according to Uber, covers him when he doesn’t have a passenger in the car. “The problem is some insurance companies are saying, ‘As soon as you log on with your app, you’re using your car commercially'” — that is, if you’re hurrying to pick someone up but don’t have a passenger yet, there’s a possibility that the driver will be covered by neither policy in an accident.


200 Accept Buyouts at Zappos After Management Changes

That went well…:

About 14 percent of the company’s workforce, or 210 out of 1,503 employees, have taken buyouts, Zappos spokeswoman Catherine Cook said.


Postgres to Support Upsert Statements

The long awaited feature is finally coming:


Syntax and usage are in the documentation.


Image Scaling Using Deep Convolutional Neural Networks

Norman Tasfi, writing for Flipboard:

This past summer I interned at Flipboard in Palo Alto, California. I worked on machine learning based problems, one of which was Image Upscaling. This post will show some preliminary results, discuss our model and its possible applications to Flipboard’s products.

Awesome results to have in mind if you ever need to upscale images.

← Previous    1    2    3    4    5       20    Next →