This blog is moving! Update your readers!
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
4y ago
Yes, I am moving onto my own website: robertgrantstats.co.uk/blog. New posts will appear there with immediate effect. Selected old posts will be migrated over the next 2 months. This site will close permanently and all content will be deleted on 31 December 2019. Why? You may have noticed the ads that appear here, and the likes and comments. These are problematic. wordpress.com is a very quick and easy way to set up a WordPress blog but it is kind of a social media platform, and that brings problems. The ads sometimes promote things inappropriate for a medical statistician’s blog, like gamblin ..read more
Visit website
Five levels of analytical automation
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
4y ago
I have been thinking more about how programming that requires minimal human input is a virtue in computer science, and hence machine learning, circles. Although there’s no doubt that is one of the central goals of programming a computer in general, I’m not convinced this extends to data analysis, which needs some thought, contextual knowledge and curation. The contextual knowledge can be broken down further into understanding where the data came from and its weaknesses, what question we are really trying to answer, and where the findings are going: the statistical literacy of the audience and ..read more
Visit website
She was like omg, and I was like no way
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
4y ago
Why are there zombie stats (they’re not true, and nobody really believes them anyway, yet they refuse to die)? I think they fulfil a cultural need, e.g. “we only use 10% of our brains” allows us to assert that we can all become smarter in various ways if we try. If you stop and think about it for even a couple of seconds, you can see that it doesn’t make sense (what’s a % of a brain? what does “use” mean? how could anyone possibly measure this?) You don’t need the pseudo-stat but it’s kind of fun; sometimes you’ll hear people saying this with different numbers. Now, if we can supply a better m ..read more
Visit website
BayesCamp prize 2019
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
Each year at the International Workshop on Computational Economics and Econometrics, I award a BayesCamp prize for the talk with clearest exposition. Last week, I discussed the strange mental contortions one can get in over voting systems. Now to announce the winner, who is… Dr Karolina Safarzynska, Associate Professor at the Faculty of Economic Sciences at the University of Warsaw Her talk was titled “a higher rebound effect under bounded rationality: interactions between car mobility and electricity generation”. This work is also in a paper forthcoming in Energy Economics. She set out a comp ..read more
Visit website
Elections, ranking and Bayes
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
I was at the International Workshop on Computational Economics and Econometrics last week, which I help to organise and is held in Rome each summer. For the second year, I’ve put up a BayesCamp prize for the presentation with the clearest exposition. Last year, I got everyone to vote at the end of the three-day workshop, but was suspicious that recall bias had pushed all the votes toward the most recent talks. Also, anyone who had to leave early didn’t get a vote. So, this time I had what appeared at first to be a Good Idea: I would put out a voting sheet at the end of each half-day session. T ..read more
Visit website
SIST thought 1: likelihood and priors for model space
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
This was a thread on Twitter but I wanted to share it here for a different audience. Today’s thought (actually 24 June but here you get free shipping) from reading Deborah Mayo’s book SIST. The likelihood principle (H1 has more evidence than H2 iff P(x|H1)>P(x|H2)) allows comparisons from pre-specified hypotheses, but not inference over the whole space of possible hypotheses and explanations. You can restrict the model and get parameter values that maximise likelihood, and that could be fine. But over the models, it gets much harder, and often , that’s what we have to do. You can maximise l ..read more
Visit website
Hand-crafted artisanal data versus autonomous machine learning
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
This was a thread on Twitter, and I thought should reach a different audience here too. If you wonder why it lacks the usual prolix rotation of the old bon mots like single malt whisky on one’s langues semblables a des langues de feu, that’s why. Article of the year: “The BS-industrial complex of phony AI” by @mikemallazzo. If you want to understand why there is an AI boom, you have to read this. I agree with 100% of it. https://gen.medium.com/the-bs-industrial-complex-of-phony-a-i-44bf1c0c60f8 (note: this is on Medium and they will snoop the hell out of you unless you go in through Tor) Here ..read more
Visit website
Can we get hyper-local infrastructure time series with StreetView and deep learning?
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
More Google StreetView ideas. Suppose you wanted a measure of infrastructure investment, or of fragility because maintenance has been cut back? How about some of those wacky instrumental variables that economists love? But you want it at hyper-local level. Well, maybe you can track this stuff visually. I picture some convolutional neural network that takes images like these, identifies common objects — road signs in this case, which are uniform and regulated so should be easy to pick out — and measures the extent of some problem — here, being overgrown with vegetation. I saw this very sign a f ..read more
Visit website
Nominalia, a region where the priests conduct Strange Rites
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
You have a nominal predictor variables with many values. That is to say, there are many categories and they do not have an innate order. Perhaps they are towns and villages in Hampshire, and you are using that alongside other data in a survey to look at the impact of perceived road surface condition on bicycle use (a causal question). Perhaps they are countries in the EU, and you want to predict voting patterns (a predictive question). They don’t have to relate to locations, though these examples do. How do you analyse them? For statisticians, they simply have to be turned into numbers somehow ..read more
Visit website
Google StreetView exercises in workshops
Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant
by Robert
5y ago
I’m now starting workshops on some topics with this small group exercise: You’re the CEO of a startup that’s going to provide house price information that’s more accurate than anything else out there. It’s going to use data analysis, which you’ve heard a lot about recently; it seems that every exciting startup is based on some kind of number-crunching. You’re the CEO, so you don’t need or want to know about the techy details. You just need the idea and then to recruit clever people to do it. And the idea is this: your software looks at Google StreetView, and because it’s previously scanned mil ..read more
Visit website

Follow Dataviz – Stats – Bayes | A blog by data sherpa Robert Grant on FeedSpot

Continue with Google
Continue with Apple
OR