It's Tuesday and that means it's time for #TidyTuesday again! ☀️ For this week's data from @nws I looked at accuracy in predicting high temperatures - finding they're often over-estimated! 🔥🔥🔥
Code: https://github.com/nrennie/tidytuesday/tree/main/2022/2022-12-20
The @R4DSCommunity welcomes you to week 51 of #TidyTuesday! We're exploring Weather Forecast Accuracy!
📁 https://bit.ly/tidyreadme
📰 https://github.com/speegled/weather_forecasts
Just moved over from twitter.
Here's my #introduction
Interested in:
#rstats #python #PyData #nlp #ml #ai #DataScience #DataAnalysis #DataViz #DataEthics #MlEthics
#criminology #policing #socialwork #MissingPersons #harm #vulnerability #SocialJustice
So I understand that the bird site has started blocking links to common/known Mastodon instances, and that doesn't seem like free speech to me. So... I did a thing.
If you go to https://spacekaren.sucks it is now a URL shortener that blocks the twitter user agent. That means you can share that link that will redirect to your Mastodon account without the pouty baby stopping you. If you find this helpful can you boost so others can find it? #TwitterMigration #Mastodon #Musk #introduction
#Tidytuesday: US Monthly State Retail Sales.
Just ran across this on the Other Site - good info about and new approaches to #parallelisation in #RStats https://www.jottr.org/2022/12/05/avoid-detectcores/
Jumping on the #introduction bandwagon! 👋
I’m Sunil, a #postdoctoral #researcher in south west #Germany working in #healthcare on #technology, #FAIR #data and #social aspects of #rare-diseases 🧬👨🏽💻🇩🇪
In my #PhD I worked with powered #wheelchair users to explore their experiences of #mobility and #access, and how #disabled people’s knowledge can inform service provision 👩🏻🦼♿️🦮🧑🦯
This spanned #disabilitystudies #sociology #urbanism #geography & #politics too! 🌆
See my work at https://scholar.google.co.uk/citations?user=oSEzsOoAAAAJ&hl=en
Calling all graduate students and early career researchers in #datascience: @rwdatasci has launched a new blog this week called DataScienceBites - and we want you to join us as contributors!
The blog publishes digestible, engaging summaries of interesting new pre-print and peer-reviewed publications in the data science space. Our goal is to make scientific papers more widely accessible.
Further details: https://realworlddatascience.net/news-and-views/editors-blog/posts/2022-12-13-dsb-launch/dsb-launch.html
What this does highlight is that MCMC _is not_ Bayesian, which some people seem to think it is. I'm a proponent of subjectivism and Bayesian probability, thus it pains me to say that frequentist statistics is useful
I *really* want to cite this paper in my thesis, but I can't see where I would put it. I'll just share it here instead. Huge fan of dogmatic papers from the world of #bayesian statistics. I don't necessarily agree with the paper, but I wish people were able to publish more stuff of this flavour
O'Hagan, Anthony. Monte Carlo is fundamentally unsound. The Statistician (1987)
@_jcken
Three sigma rule should be quite safe unless prior odds on H1 are low
https://royalsocietypublishing.org/doi/10.1098/rsos.140216
and
https://www.tandfonline.com/doi/full/10.1080/00031305.2018.1529622
I'm excited to welcome you all to week 50 of #TidyTuesday from the
@R4DSCommunity! This week we're going shopping!
📂 bit.ly/tidyreadme
📰 census.gov/retail/state_retail_sales.html
#tidytuesday week 50: US retail sales 💳
There was a 7 fold increase in clothing sales in 2021
This week's #TidyTuesday data is from the US Census Bureau and looks at changes in retail sales values over the last couple of years! 💵 Found some interesting (inverse) patterns between food/beverage and gas station sales...
Code: https://github.com/nrennie/tidytuesday/tree/main/2022/2022-12-13
did some actual statistics (!!) with #TidyTuesday this week. It doesn't seem to be very well known, but Pukelsheim's 3 sigma rule allows you to perform robust inferences with very few assumptions. Here it detects the impact of covid lockdowns and whatever else happened in 2021. The model is an AR(1) model which uses data up to time t to predict what happens at t+1. No predictions made for first 12 time points
Code: https://github.com/jcken95/tidytuesday/tree/main/2022/2022-12-13
Paper on 3 sigma rule (paywall?): https://jstor.org/stable/2684253
I decided to have a little chat with #ChatGPT. Maybe it was the nature of the questions I was asking, but the responses were clear about the limitations of the model and its outputs. I was also pretty impressed with the way it summarised our "conversation" when asked to do so.
As a #journalist and #editor, I can see this sort of text summarisation as being a quite helpful tool when working with, say, a 10,000-word interview transcript
I have a stats PhD from Newcastle and work as a senior data scientist at UKHSA
Royal statistical society volunteer
🌲 🐕 🚵♂️ | views own | he/him