Peder R. Braadland
LightNews LightNews
banner
braadland.bsky.social
Peder R. Braadland
@braadland.bsky.social
670 followers 830 following 1.6K posts
https://peder.quarto.pub/blog
Father x3. Rstats & data visualisation

Currently studying gut bacterial metabolism in primary sclerosing cholangitis
Norwegian green party 🌱
🇳🇴 🇩🇪 🏴󠁧󠁢󠁥󠁮󠁧󠁿
Posts Replies Media Videos
braadland.bsky.social
Litt lenger vest, se under.

Har vært lite der de siste årene, men kan ikke huske at jeg så den sist jeg var der, men mulig jeg tar feil
Utsnitt fra Google maps som viser hvor sykkelparkering var/er
October 13, 2025 at 6:36 PM
braadland.bsky.social
September 24, 2025 at 6:29 AM
braadland.bsky.social
Hehe 😜
September 8, 2025 at 8:58 AM
braadland.bsky.social
I tilfelle noen fortsatt ikke har bestemt seg #mdg
MDG-brosjyre laget av niåringen
September 6, 2025 at 12:19 PM
braadland.bsky.social
Til tross for at halvparten av Høyrevelgerne i Oslo er for eiendomsskatt, vil Høyre altså fjerne eiendomsskatten. Samtidig er det slik at de som får mest skattekutt ved kutt i eiendomsskatten typisk er Høyre-velgere på Oslo Vest (riktignok data fra Kommunevalget 2023).
Sammenheng mellom andel (per valgkrets) som stemmer Høyre og gjennomsnittlig betalt eiendomsskatt. Man kan se at mennesker som bor i valgkretser med høy gjennomsnittlig eiendomsskatt også i større grad stemmer Høyre.
September 3, 2025 at 11:56 AM
braadland.bsky.social
Sånn ser det forresten ut hvis man zoomer litt ut. Tallene er fra pollofpolls.no forresten
September 2, 2025 at 8:19 PM
braadland.bsky.social
Utfra min egne kjappe analyse, står ditt punkt 1 seg også om man bruker gjennomsnitt for valgkampenes siste 14 dager ved siste tre stortingsvalg. (alt som havner under diagonalen hinter om underprestering ift meningsmålingene; "dårligere resultat enn man skulle tro")
"Kalibrering" av meningsmålinger (siste to uker før valgdagen) vs faktiske valgresultater ved siste tre stortingsvalg i Norge
September 2, 2025 at 8:18 PM
braadland.bsky.social
Selv om den støtt og stadig postes her, synes jeg den er så passende at jeg poster den en gang til
Yes, the planet got destroyed, but for a beautiful moment in time we created a lot of value for shareholders - meme
September 2, 2025 at 7:54 PM
braadland.bsky.social
Fra NRKs forside
Bilde fra NRKs forside: Sp og MDG: Bommet grovt om rovdyr-motstand
September 1, 2025 at 7:41 AM
braadland.bsky.social
Går jo i riktig retning dette, så her er det bare å holde dampen oppe 🟩
Polling data for MDG (Norwegian green party) over the past few months. A loess fit with 95% confidence interval is shown, alongside individual polling results colored by pollster. The 4% threshold is emphasized.
August 22, 2025 at 8:26 PM
braadland.bsky.social
(PS: Det er samvariasjon også om man ser på bydelene med gjennomsnittlig betalt eiendomsskatt under 1000/år)
Viser sammenhengen mellom gjennomsnittlig betalt eiendomsskatt og andel stemmer til Høyre, i følge beregninger gjort av Frifagbevegelse. Se https://frifagbevegelse.no/nyheter/hoyrevelgere-tjener-mest-pa-kutt-i-eiendomsskatten-6.158.1081676.848e631ab0
May 22, 2025 at 3:48 PM
braadland.bsky.social
Vet ikke hvem som trenger dette, men her er altså en sammenlikning av uvektede snitt av meningsmålingsdata fra @pollofpolls.no for de siste to ukene før de tre siste Stortingsvalgene, og faktiske valgresultater. Overraskende god "kalibrering" her? #ggplot2 #rstats
peder.quarto.pub/blog/posts/a...
Shows the actual election results vs predicted (unweighted average of polls for final two weeks leading up to the election) for the parliamentary elections in Norway for 2013, 2017 and 2021. There is an excellent concordance between the two, indicating that the polls accurately predict election outcomes.
May 13, 2025 at 9:32 AM
braadland.bsky.social
This is such a bad idea #stats
Screen Capture of email from Biorender advertising for their new feature allowing users to upload spreadsheets to have the software automatically produce data.visualizations
May 10, 2025 at 6:01 AM
braadland.bsky.social
Not sure if youre looking for how to make them but @mjskay.com's ggdist::stat_interval does the job
mjskay.r-universe.dev/articles/ggd...
April 30, 2025 at 5:56 PM
braadland.bsky.social
A non-conventionally plotted Kaplan-Meier plot using data from a 1994 PBC trial of UDCA vs placebo. Previous post wrongly said endpoints were only transplant or death (made outcomes seem overly dim) - in reality, endpoint also included treatment failure 📊

#30daychartchallenge #rstats #ggplot2
Kaplan Meier plot showing probabilities of treatment failure for patients with primary biliary cholangitis (PBC) given UDCA or placebo in an RCT from the 1990's. The plot indicates that UDCA was superior to placebo in delaying treatment failure, which is a composite endpoint that was defined as death, liver transplant, or a set of indices of disease progression.
April 25, 2025 at 1:50 PM
braadland.bsky.social
Agree it's busy with all the parties! I originally made this plot as a way to compare how we perceive scattered points vs confidence intervals (plot below). I think the scattered points communicate a more nuanced story. So many ways to smooth which has a big impact on how people interpret the plot!
April 21, 2025 at 5:59 AM
braadland.bsky.social
For day 19 (smooth) of the #30daychartchallenge I show some long-term political party polling data (pollofpolls.no) from Norway (Stortingsvalg). Made with #rstats #ggplot2. The viz is probably helpful to appreciate long-term trends but not so much recent developments
peder.quarto.pub/blog/posts/a...
Polling data for individual political parties for the Norwegian parliamentary elections (every four years; "Stortingsvalg"). The polling data were downloaded from PollOfPolls.no on April 20th 2025
April 20, 2025 at 7:47 PM
braadland.bsky.social
Looking at the major threats birds face today. Remember seeing a plot dealing w/"overflow" ike this, but can't remember where. Anyways, here's day 17 of the #30daychartchallenge 99% #rstats #ggplot2 📊
Perhaps not ideal for making friends with cat people 😄

peder.quarto.pub/blog/posts/a...
Shows a bar plot of top killers of birds (annually for the US). Cats are the major killer, at 2.4 billion, followed by crashing into building windows (600 million), vehicles at 214 million and so on. Wind turbines are estimated to cause only around 230 thousand bird deaths annually.
April 17, 2025 at 4:49 PM
braadland.bsky.social
Actually not sure if they are called that. Thinking of it, having all countries in the same plot would lead to them all overlapping in the first period. Could facet by country which would emphasize the downward trend in all countries
April 16, 2025 at 1:38 PM
braadland.bsky.social
#30daychartchallenge 📊 Day 15 Complicated. Using simulations to plot and "prove" the surprising birthday paradox in #rstats #ggplot2. Why surprising? You only need to gather *23* random people for P(two or more shared birthdays) to exceed 50%!

peder.quarto.pub/blog/posts/a...
Shows the association between number of (random) people in a room and the probability that two or more of them share the same birthday. The data were generated using simulations in R. Each dot represents the probability of two or more people having the same birthday in rooms with various number of people. The probabilities were calculated as the fraction of simulated cases where two or more people shared a birthday. The simulation was repeated 10 times for each room size, giving the distribution of probabilities. The birthday paradox is that only 23 people are needed for the probability of two or more of them sharing a birthday to exceed 50%!
April 15, 2025 at 7:05 PM
braadland.bsky.social
Late day 11 of #30daychartchallenge. #rstats #ggplot2 climate stripes 📊 with a twist or two. Northern Hemisphere data. Code coming to peder.quarto.pub/blog/posts/a...
Climate stripes from 1850 to 2025, faceted over four rows.
April 14, 2025 at 1:07 PM
braadland.bsky.social
😄 That's fine. Thanks to your graph I spent way too much time trying to reproduce it in #rstats. Now back to work!
df %>% ggplot(aes(x = date, y = News_Based_Policy_Uncert_Index))+ geom_hline(yintercept = seq(200, 600, 200), color = "#e6dace", size = 0.15)+ geom_line(color = "#d14600", linewidth = 0.6) + geom_point(aes(x = max(date), y = 1000), color = "#d14600")+ theme_minimal(base_family = "Roboto", base_size = 9)+ theme( panel.grid.major = element_blank(), panel.grid.minor = element_blank(), axis.line.x = element_line(color="#5e5250", size = 0.4), axis.ticks.x = element_line(color="#5e5250", size = 0.4), axis.text = element_text(color = "#5e5250"), axis.title = element_blank(), plot.title = element_text(vjust = -21, color = "#332f2c", size = 13), plot.subtitle = element_text(vjust = -40, color = "#5e5250", size = 7), plot.title.position = "plot", plot.background = element_rect(fill = "#fff1e6", color = NA), plot.caption = element_text(size = 5, color = "#5e5250", hjust = 0), plot.caption.position = "plot", plot.margin = unit(c(-7, 1, 1, 1), "mm") ) + scale_y_continuous(breaks = seq(0, 600, 200))+ labs(title = "Uncertainty has exploded under Trump", subtitle = "US Economic Policy Uncertainty Index", caption = "Source: policyuncertainty.com/\nBased on @jburnmurdoch.ft.com's graphic\nPeder Braadland")+ scale_x_date(expand = c(0.015, 0.015))+ annotate(geom = "text", label = "Trump", x = as.Date("2024-10-01"), y = 1000, vjust = 0.5, size = 2, color = "#5e5250", hjust = 1, family = "Roboto")+ annotate(geom = "text", label = "Covid-19\npandemic", x = as.Date("2019-11-01"), y = 460, vjust = 0.5, lineheight = 0.85, size = 2, color = "#5e5250", hjust = 1, family = "Roboto")
April 11, 2025 at 12:18 PM
braadland.bsky.social
I couldnt resist 😆 Code in alt text. Nothing fancy, just some simple "hacks" like assigning breaks on y-axis and creating "fake" title and subtitle using annotate() #ggplot2 #rstats

Didn't have the most updated data so I just added another point for April.
df <- read_delim("US_Policy_Uncertainty_Data.txt") p <- df %>% mutate( Year = as.integer(Year), date = ymd(paste(Year, Month, "01", sep = "-")) ) %>% filter(!is.na(Year)) %>% filter(Year > 2004) %>% bind_rows(tibble(Year = 2025, Month = 4, Three_Component_Index = NA, News_Based_Policy_Uncert_Index = 1000, date = as.Date("2025-04-01"))) %>% ggplot(aes(x = date, y = News_Based_Policy_Uncert_Index))+ geom_hline(yintercept = seq(200, 600, 200), color = "#e6dace", size = 0.15)+ geom_line(color = "#d14600", linewidth = 0.6) + geom_point(aes(x = max(date), y = 1000), color = "#d14600")+ theme_minimal(base_family = "IBM Plex Sans", base_size = 9)+ theme( panel.grid.major = element_blank(), panel.grid.minor = element_blank(), axis.line.x = element_line(color="#5e5250", size = 0.4), axis.ticks.x = element_line(color="#5e5250", size = 0.4), axis.text = element_text(color = "#5e5250"), axis.title = element_blank(), plot.background = element_rect(fill = "#f7f1f0", color = NA), plot.caption = element_text(size = 5, color = "#CCCCCC" ) ) + scale_y_continuous(breaks = seq(200, 600, 200))+ annotate(geom = "text", x = as.Date("2005-01-01"), y = 840, label = "Uncertainty goes boom", hjust = 0.13, color = "#332f2c", size = 5, family = "IBM Plex Sans")+ annotate(geom = "text", x = as.Date("2005-01-01"), y = 700, label = "US Economic Policy Uncertainty Index", hjust = 0.13, color = "#5e5250", size = 3, family = "IBM Plex Sans")+ annotate(geom = "text", x = as.Date("2024-10-01"), y = 1000, hjust = 1, label = "Dump", color = "#5e5250", family = "IBM Plex Sans", size = 2)+ coord_cartesian(clip = "off")+ scale_x_date(expand = c(0.015, 0.015))+ labs(caption = "Source: policyuncertainty.com/, based on @jburnmurdoch.ft.com's graphic" ) ggsave("p.jpg", p, width = 70, height = 60, unit = "mm", dpi = 500)
April 11, 2025 at 9:28 AM
braadland.bsky.social
#30daychartchallenge playing around with "back-to-back" histograms using #rstats ggdist::stat_histinterval() #ggplot2 #dataviz

peder.quarto.pub/blog/posts/a...
Back-to-back histograms indicating distributions of run lengths from before 2020 and after 2020. The median and interquartile ranges are shown. Plot was made using ggplot2 in R with the ggdist package.
April 10, 2025 at 7:41 AM
braadland.bsky.social
In your example you dont let z predict x, but you allow x to predict z, right?

So in my example, I'm clear of committing the crime of allowing ID to predict age and sex?
April 8, 2025 at 1:17 PM