I had this tweet appear in my timeline Okay, the top answer to this turns out to be incredibly interesting: https://t.co/rcMcmTEQ87 pic.twitter.com/Pl8wFfgOfy — Clive Thompson (@pomeranian99) August 7, 2017 Which links through to a quora discussion about the internet formulation of z/(x+y) + y/(x+z) + x/(y+z) = 4 and an example image of With the… Continue reading Making a problem harder to make it easier
Using base R (as I did previously) means thinking about what do I want plotted where. Using ggplot means thinking about what job is the data doing in this graph. This is just running through what I was thinking when making my own version the Scientific American births per year style graph, but using… Continue reading Circular banded graphs for ggplot
Since I had previously played with NZ birth data, so had a bunch of code lying around New Zealand Births and Friday the 13th and Scientific America published some visualisations https://blogs.scientificamerican.com/sa-visual/why-are-so-many-babies-born-around-8-00-a-m/ I thought I would match the time of year visualisation with the New Zealand data Because New Zealand has opposite seasons, but a seasonality that… Continue reading Births again
I just released what I would call an important project I did in my spare time. Overturns scientific consensus. 7+ Sigma. Easy stats. Public data. Reproducible. Replicable. https://thoughtfulbloke.github.io/eqnz
I am of the opinion that confidence intervals should be shown as a range, and coloured/ shaded bands are a good way to show it. They are a probability distribution after all (but you can’t make histogram because it disappears to invisibility after 3 sigma, so bands seems the best plan). Admittedly, I am trying… Continue reading A modest proposal- coloured confidence intervals
In theory, R is very comfortable with unicode (individual operating systems and locales may vary this) so it should be possible to type in unicode and graph in unicode. But the process is limited by the palettes of the default fonts. So if you see what symbols will actually plot, the results can be disappointing.… Continue reading Getting the cool symbols onto graphs in R
Carrot prices http://www.stuff.co.nz/business/90367412/a-kilogram-of-carrots-costs-over-a-dollar-more-than-last-year Looking at both the raw (so to speak) and seasonally adjusted figures, it looks to me like carrots, as the most extreme difference, got a discussion in the press release, but there is more of a steady upward creep in prices generally. The released figures go to the Fruit and Vegetable level,… Continue reading What is orange and been in the news
This is just showing how to process the New Zealand baby names spreadsheet provided by the New Zealand Governemnt, and also make streamgraphs, available from github with devtools::install_github(“hrbrmstr/streamgraph”) download.file(“https://smartstart.services.govt.nz/assets/files/Top-baby-names-1954-2016.xlsx”, destfile=”babynames.xlsx”, mode=”wb”, method=”libcurl”) library(readxl) library(dplyr) library(streamgraph) female <- read_excel(“~/Desktop/babynames.xlsx”, sheet = “Girls’ Names”, col_names = FALSE, skip = 7) female <- female[1:100,] male <- read_excel(“~/Desktop/babynames.xlsx”, sheet… Continue reading New Zealand Names
I decided to check world usage of R. After a few false starts (cran download logs don’t include IP blocks) I settled on using Google Trends to find the frequency of the search term R Data and downloaded the interest by region table. But being from tiny New Zealand, I am also interested in a… Continue reading WheRe in the woRld is R
Sometimes I get the problems no-one else has a solution for, particular those involving data transformation. Qualtrics (online survey tool) lets you create random selection questions, where you provide a pool of answer options and set Qualtrics to randomly display a specified number of the answer options to the person doing the survey. Which can… Continue reading A fix for Qualtrics random selection data
thoughtfulbloke aka David Hood 