David Hood
@thoughtfulnz.bsky.social
I like collating data,
and getting thoughts in the brain.
I run https://thoughtfulnz.quarto.pub/nzcovidreport/
All for Public data, Aotearoa New Zealand, and informed communities. #openData #NewZealand #Aotearoa #rstats. He/him.
He tangata tiriti ahau
and getting thoughts in the brain.
I run https://thoughtfulnz.quarto.pub/nzcovidreport/
All for Public data, Aotearoa New Zealand, and informed communities. #openData #NewZealand #Aotearoa #rstats. He/him.
He tangata tiriti ahau
#30DayMapChallenge Day 11. A minimal map.
I assume enough of my regular readers could place themselves, so it is a map. Made in Inkscape.
I assume enough of my regular readers could place themselves, so it is a map. Made in Inkscape.
November 10, 2025 at 5:57 PM
#30DayMapChallenge Day 11. A minimal map.
I assume enough of my regular readers could place themselves, so it is a map. Made in Inkscape.
I assume enough of my regular readers could place themselves, so it is a map. Made in Inkscape.
Where I am, I'm favour of directly calling out the values framing misinformation, to use the disjoint from reality as lever to draw attention to motives. For instance, recently right wing politicians here expressed imported concern about fertility rates (great replacement theory adjacent concerns)
November 9, 2025 at 8:09 PM
Where I am, I'm favour of directly calling out the values framing misinformation, to use the disjoint from reality as lever to draw attention to motives. For instance, recently right wing politicians here expressed imported concern about fertility rates (great replacement theory adjacent concerns)
something something for something something
November 9, 2025 at 4:35 AM
something something for something something
It felt to me like #30DayMapChallenge on Bluesky is a lot busier this year than last, so I used the API (atrrr in #rstats) to count up how many posts since the month started. Yup, I was remembering correctly.
November 8, 2025 at 5:34 AM
It felt to me like #30DayMapChallenge on Bluesky is a lot busier this year than last, so I used the API (atrrr in #rstats) to count up how many posts since the month started. Yup, I was remembering correctly.
#30DayMapchallenge Day 8, Urban.
People don't realise what a big deal it is for acceptance of place names to change in such a short time without wars/dictators/invasion/etc. Just people, within a generation, deciding it is the right thing to do. mapdata::NZHires baseman + Affinity Publisher.
People don't realise what a big deal it is for acceptance of place names to change in such a short time without wars/dictators/invasion/etc. Just people, within a generation, deciding it is the right thing to do. mapdata::NZHires baseman + Affinity Publisher.
November 8, 2025 at 12:53 AM
#30DayMapchallenge Day 8, Urban.
People don't realise what a big deal it is for acceptance of place names to change in such a short time without wars/dictators/invasion/etc. Just people, within a generation, deciding it is the right thing to do. mapdata::NZHires baseman + Affinity Publisher.
People don't realise what a big deal it is for acceptance of place names to change in such a short time without wars/dictators/invasion/etc. Just people, within a generation, deciding it is the right thing to do. mapdata::NZHires baseman + Affinity Publisher.
I have been checking how to create Gastner-Newman cartograms (the ones where the country sizes are rescaled by a value)
November 7, 2025 at 1:03 AM
I have been checking how to create Gastner-Newman cartograms (the ones where the country sizes are rescaled by a value)
For the #30DayMapChallenge day 7 Accessibility, in 2021 to help someone looking at health equity and the vaccine rollout, I worked out the road travel distance from every SA2 centre point to every hospital using purely open sources so that it could be freely used. #rstats code in alt text.
November 7, 2025 at 12:10 AM
For the #30DayMapChallenge day 7 Accessibility, in 2021 to help someone looking at health equity and the vaccine rollout, I worked out the road travel distance from every SA2 centre point to every hospital using purely open sources so that it could be freely used. #rstats code in alt text.
A numeracy comment of mine is that births only strongly correlate to changes in the school aged population to the extent that a country is too unpleasant to be a migration destination for families. We get people in my country buying into imported framing without considering local evidence.
November 6, 2025 at 5:38 PM
A numeracy comment of mine is that births only strongly correlate to changes in the school aged population to the extent that a country is too unpleasant to be a migration destination for families. We get people in my country buying into imported framing without considering local evidence.
(a few hours later) Oh, come off it Rakiura
November 6, 2025 at 8:24 AM
(a few hours later) Oh, come off it Rakiura
NZ and AU accounts, accustomed to a more vigorous use of language as punctuation than the comparatively prim US, might at the very least set it to warn so you know people are saying things even if you are nervous about be unshielded from "no constructive purpose". Wordcloud of NZ Twitter 2021
November 6, 2025 at 7:00 AM
NZ and AU accounts, accustomed to a more vigorous use of language as punctuation than the comparatively prim US, might at the very least set it to warn so you know people are saying things even if you are nervous about be unshielded from "no constructive purpose". Wordcloud of NZ Twitter 2021
For #30DayMapChallenge day 5: Earth, I wasn't particularly motivated until I remembered that one guy in Auckland (AoNZ will understand what I mean) so learned to process geojson- Earthquakes (+) over 3.5 Magnitude, south of -38.8 and where and how strongly they were felt. #rstats code in alt text
![library(readr)
library(curl)
library(geojsonio)
library(dplyr)
library(tidyr)
library(ggplot2)
library(mapdata)
# quakes > 3.5, south of -38.8 Lat
searcheq <- "https://quakesearch.geonet.org.nz/csv?bbox=165.4794,-49.4789,179.4093,-38.8&minmag=3.5&startdate=2025-09-06T0:00:00&enddate=2025-11-06T1:00:00"
nzeqs <- read_csv(searcheq)
write.csv(nzeqs,"nzeqs.csv")
dir.create("eqdata")
#as geojsonio seems to expect a suffix, I am making a local cache
for(public_id in nzeqs$publicid){
urlloc = paste0("https://api.geonet.org.nz/intensity?type=reported&publicID=", public_id)
local_url <- paste0("eqdata/", public_id, ".geojson")
handle <- new_handle()
curl_download(urlloc, local_url, handle = handle)
}
all20 <- list.files("eqdata", full.names = TRUE, pattern=".*geojson$")
extractdata <- function(x){
#Some had no felt reports, so using tryCatch
tryCatch(
{
felt_data <- geojson_read(x, what = "sp")
},
error = function(e) {
return(NULL)}
)
feltcoords <- as.data.frame(felt_data@coords) |> rename(long=1,lat=2)
feltcoords$mmi <- felt_data@data[["mmi"]]
feltcoords$mmicount <- felt_data@data[["count"]]
return(feltcoords)
}
allfelts <- bind_rows(lapply(all20,extractdata))
expanded_felts <- allfelts |>
uncount(weights = mmicount) |>
arrange(mmi)
nzoutline <- map_data("nzHires")
ggplot() +
geom_polygon(data=nzoutline, aes(x=long, y=lat, group = group), colour="white", fill="white")+
geom_jitter(data=expanded_felts, aes(x=long,y=lat, colour=mmi))+
geom_point(data=nzeqs, aes(x=longitude,y=latitude), pch="+", size=8)+
annotate("text", y=-38.6,x=170, label="-38.8")+
scale_colour_viridis_c(name="Felt\nIntensity")+
geom_hline(yintercept = -38.8)+
coord_quickmap(xlim=c(166,178), ylim=c(-47.5,-34.5)) +
labs(title = "#30DayMapChallenge 2025 Day 05: Earth",
subtitle = "Felt reports of earthquakes(+) > 3.5, south of -38.8 in the last two months",
caption = "Source: Geonet") +
theme_void(paper="grey")+
theme(legend.position = c(0.8, 0.2))](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:c25e6vi7ojpyitmoi3xpuvpf/bafkreifxn5y4enweiua7lg7725ibb3dompyggtutddwn6zei5zq3cdfp4y@jpeg)
November 6, 2025 at 2:20 AM
For #30DayMapChallenge day 5: Earth, I wasn't particularly motivated until I remembered that one guy in Auckland (AoNZ will understand what I mean) so learned to process geojson- Earthquakes (+) over 3.5 Magnitude, south of -38.8 and where and how strongly they were felt. #rstats code in alt text
For #30DayMapChallenge day 4, My Data. I haven't been everywhere, but I have been to 73 of the 116 locations in the song "I've been everywhere" (allowing for some questionable location names in the song).
November 5, 2025 at 2:49 AM
For #30DayMapChallenge day 4, My Data. I haven't been everywhere, but I have been to 73 of the 116 locations in the song "I've been everywhere" (allowing for some questionable location names in the song).
Here is a version that rearranges the legend order and swaps the don't know pattern from "plasma" to "circle" for stronger contrast
November 4, 2025 at 9:27 PM
Here is a version that rearranges the legend order and swaps the don't know pattern from "plasma" to "circle" for stronger contrast
A problem is that "trust" is a pretty conditional term that when used in its broadest is a bit difficult to say what is being described by the participants. So I think you need to contextualised with other institutional expertise trust related questions like this. But, broadly, NZ pro-science
November 3, 2025 at 11:19 PM
A problem is that "trust" is a pretty conditional term that when used in its broadest is a bit difficult to say what is being described by the participants. So I think you need to contextualised with other institutional expertise trust related questions like this. But, broadly, NZ pro-science
A bit more tweaking- using ggpattern (and alpha to mute it on the non-Don't Knows) to make it a bit clearer than for these purposes not giving a definitive answer is considered a different thing, and tweaking the Trust level factor/colour order.
A bit of the key #rstats graph making code in alt text
A bit of the key #rstats graph making code in alt text
![prosci <- sav23 |>
mutate(trust = case_when(A11d == "9. Don’t know" ~ "Don't Know/\nDidn't Answer",
is.na(A11d) ~ "Don't Know/\nDidn't Answer",
TRUE ~ gsub(".*\\. ","",A11d))) |>
summarise(.by=c(mvpartyvote, trust),
weighted = sum(vwgt)) |>
group_by(mvpartyvote) |>
mutate(percentage= 100*weighted/sum(weighted),
percent= ifelse(trust == "Don't Know/\nDidn't Answer", percentage * -1, percentage),
party = ifelse(is.na(mvpartyvote),"unknown", gsub(".*\\. ","",mvpartyvote))) |>
ungroup()
partorder <- prosci |>
filter(trust %in% c("Trust a lot", "Trust somewhat")) |>
summarise(.by=party,
orderscore = sum(percent)) |>
arrange(desc(orderscore))
vpal <- viridis_pal(alpha = 1, begin = 0, end = .9, direction = -1, option = "C")(4)
vcol <- c("#EEEEEE",vpal)
factrst <- c("Don't Know/\nDidn't Answer", "Trust a lot", "Trust somewhat", "Don’t trust very much", "Don’t trust at all")
prosci |>
mutate(party = factor(party, levels=partorder$party),
trust = factor(trust, levels = factrst[c(1,5,4,3,2)])) |>
ggplot(aes(x=party,y=percent, fill=trust)) +
# geom_col(colour="black", linewidth=0.05) +
geom_col_pattern(aes(fill = trust,
pattern_alpha = trust),
colour="black",linewidth=0.05,
pattern = 'plasma',
pattern_fill = "grey0") +
theme_david() +
scale_fill_manual(name="Trust level", values=vcol) +
scale_pattern_alpha_manual(name="Trust level", values = c(.5,0,0,0,0)) +
scale_y_continuous(breaks=c(-15,0,25,50,75,100),
labels=c(15,0,25,50,75,100)) +
labs(title="NZES 2023: How much trust do you have in scientists? by party vote",
subtitle="In order of trust, most trusting on left. Weighted percent.",
y="Percent of voters", x="Party voted for",
caption="Source: 2023 NZES")+
theme(axis.line.y = element_blank(), axis.line.x = element_blank())](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:c25e6vi7ojpyitmoi3xpuvpf/bafkreieack4v2snbnw5gecbtgxf5f6g2iysdkmegvaj5dzxjwagr62ygsm@jpeg)
November 3, 2025 at 10:33 PM
A bit more tweaking- using ggpattern (and alpha to mute it on the non-Don't Knows) to make it a bit clearer than for these purposes not giving a definitive answer is considered a different thing, and tweaking the Trust level factor/colour order.
A bit of the key #rstats graph making code in alt text
A bit of the key #rstats graph making code in alt text
Just trying out a way of showing Don't knows/Didn't answers.
Calculating the percents of Trust level and Party, then multiplying the don't know by -1 and using geom_col()
#rstats
Calculating the percents of Trust level and Party, then multiplying the don't know by -1 and using geom_col()
#rstats
November 3, 2025 at 9:25 AM
Just trying out a way of showing Don't knows/Didn't answers.
Calculating the percents of Trust level and Party, then multiplying the don't know by -1 and using geom_col()
#rstats
Calculating the percents of Trust level and Party, then multiplying the don't know by -1 and using geom_col()
#rstats
The #30DayMapChallenge day 2, lines. The route of verse 4 of the NZ version of "I've been everywhere" (Adapted by John Grenell (John Hore), from the original Australian
version composed by Geoff Mack) #rstats code in alt text (just)
version composed by Geoff Mack) #rstats code in alt text (just)
![# Placenames: https://gazetteer.linz.govt.nz
# lyrics: https://www.ivebeeneverywhere.com.au/the-song/song-versions/new-zealand-version/
verse4 <- c("Woodville", "Dargaville", "Lumsden", "Katikati", "Naseby", "Cambridge", "Porirua", "Mararoa", "Hastings", "Tikitiki", "Tauranga", "Auckland", "Naenae", "Waitaha", "Hamilton", "Poroporo", "Taupō", "Timaru", "Oamaru", "Tihoi", "Awanui", "Wanganui", "Pauanui")
library(dplyr)
library(ggplot2)
library(mapdata)
nzoutline <- map_data("nzHires")
gaz <- read.csv("gaz_csv.csv")
# 1 location can have had both official & unofficial (past) entries, the arrange/group/slice gets the most official.
# Also massaging unofficial names
where <- gaz[gaz$name %in% c(verse4,"Mararoa River"),] |>
arrange(name, status) |>
group_by(name) |>
slice(1) |>
ungroup()
where$name[where$name == "Mararoa River"] <- "Mararoa"
where$crd_longitude[where$name == "Waitaha"] <- 360 - 176.5817
where <- where[,c("name","crd_latitude", "crd_longitude")] |> rename(verse4=name)
forgraph <- data.frame(verse4,ordr=1:23) |>
inner_join(where,by = join_by(verse4)) |>
mutate(xto=lag(crd_longitude),yto=lag(crd_latitude))
ggplot() +
geom_polygon(data=nzoutline, aes(x=long, y=lat, group = group), colour="white", fill="white")+
geom_curve(data=forgraph, aes(x=crd_longitude,y=crd_latitude,xend=xto,yend=yto)) +
theme_void(paper="grey") +
coord_quickmap(xlim=c(130,NA)) +
labs(title = "#30DayMapChallenge 2025 Day 02: Lines",
subtitle = "Route of verse 4 of 'I've Been Everywhere' (NZ version)\n",
caption = "Source: NZ Placename Gazette, Toitū Te Whenua Land Information New Zealand, CC 4.0\nLyrics: https://www.ivebeeneverywhere.com.au/the-song/song-versions/new-zealand-version/") +
annotate("text", x=135, y=-30, hjust=0,vjust=1, label="Woodville Dargaville\nLumsden Katikati\nNaseby Cambridge\nPorirua Mararoa\nHastings Tikitiki\nTauranga Auckland\nNaenae Waitaha\nHamilton Poroporo\nTaupo Timaru\nOamaru Tihoi\nAwanui Wanganui\nPauanui")](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:c25e6vi7ojpyitmoi3xpuvpf/bafkreih7ily2gja2ilxvyqmvy6rcgskizy7oj3kxqgi2naono4gmunskaq@jpeg)
November 2, 2025 at 5:01 AM
The #30DayMapChallenge day 2, lines. The route of verse 4 of the NZ version of "I've been everywhere" (Adapted by John Grenell (John Hore), from the original Australian
version composed by Geoff Mack) #rstats code in alt text (just)
version composed by Geoff Mack) #rstats code in alt text (just)
As it happens, my office is in a building that was formerly a data centre bridge.
November 1, 2025 at 9:35 PM
As it happens, my office is in a building that was formerly a data centre bridge.
As it is November 1st in NZ, here is Day 1 of the #30DayMapChallenge - Points. In this case high points in my part of Aotearoa. Every officially gazetted place name of type hill (which includes mountains). #rstats code that made it in alt text. Used geom_text() to plot unicode symbols, as one does.
October 31, 2025 at 6:36 PM
As it is November 1st in NZ, here is Day 1 of the #30DayMapChallenge - Points. In this case high points in my part of Aotearoa. Every officially gazetted place name of type hill (which includes mountains). #rstats code that made it in alt text. Used geom_text() to plot unicode symbols, as one does.