Thursday, 24 November 2016

Cute Cat Thursday 2016-11-24: Joni

One day early in honour of Thanksgiving. If you celebrate it, may your holiday include good food, good company, and relaxing naps.


Friday, 18 November 2016

Polls, Predictions, and Russian Roulette

In the aftermath of the US election, prediction websites have been heavily criticised for giving Clinton a high chance of victory. This is not the fault of the sites themselves; I think it has more to do with the mindset of the users.

This is really safe, right?

Cute Cat Friday 2016-11-18: Joni's Gotcha Day

Today is a very special Cute Cat Friday: It marks exactly one year since we brought Joni home from the Blue Cross.

Sadly things didn't work out between her and our other cat Belle, and we decided to rehome Belle. It was hard, but we are more sure than ever that we made the right choice. Joni has thrived in our home. She is amazingly patient with our toddler, maybe because she is a mother herself and thinks of him as an unusually large kitten.

Joni is a happy, confident, affectionate cat and a much loved member of our family. Here's to many more years together.




Friday, 11 November 2016

The Wolf Hall President

Donald Trump will be President of the United States.

I feel physically sick typing those words, but this is the new reality and we have to deal with it. The fact of Trump's victory is bad enough, but more sickening is the rush to normalise him.

Donald Trump.
Source: Wikimedia Commons

Cute Cat Friday 2016-11-11: Joni

Human governments come and go, but the cat abides.


Tuesday, 8 November 2016

Reasons To Be Afraid

First things first: There is no reason to panic.

Although FiveThirtyEight give Trump a 28% chance of winning the election, they are very much an outlier among prediciton sites. I prefer the Princeton Election Consortium (PEC), who give Trump a chance of less than 1% to win the popular vote, and just over 5% to win the electoral college (and with it the presidency).

It's not just wishful thinking. With my professional data-analyst hat on, I find this Huffington Post critique of 538's model results to be persuasive. So far as we can tell, 538 have an overcomplicated model which assigns far too much probability to weird outlier scenarios. Unlike 538, PEC make their data and methods fully public; I haven't delved into their code, but their general methodology appears sound.

PEC predict this election map, very likely a comfortable victory for Clinton:

Princeton Election Consortium map at 270toWin
Click the map to create your own at 270toWin.com