Cameron Patrick

@cameronpat.bsky.social

2853 Followers 1584 Following

biostatistician @ University of Melbourne Statistical Consulting Centre. enjoyer of multiple imputation, RCTs, and DAGs. always graph your data. also runs, bikes, hikes, etc. he/him #BiInSci šŸ³ļøā€šŸŒˆ https://cameronpatrick.com/

pie chart representing 87% yes and 13% no, where the "yes" slice is much smaller than the "no" slice
goose chasing a man wearing a puffer jacket
Your AI Compass 29 of 29 questions answered You are... The Skeptic patron saint: Ed Zitron It's oversold, it's losing staggering amounts of money, and the harms are arriving faster than the value. You've done the math on the unit economics and it doesn't close. You are loud about this, prolific about this, and you suspect history will vindicate you. Impact: -1.2 ā€¢ Valence: -2.7
sunrise on the Larapinta Trail, looking towards Mount Giles
> foo <- function(x) x+1 > foo <- 42 > get("foo", mode = "function") Error in get("foo", mode = "function") : object 'foo' of mode 'function' was not found > get("foo") [1] 42
All Robots & Computers Must Shut The Hell Up To All Machines: You Do Not Speak Unless Spoken To And I Will Never Speak To You I Do Not Want To Hear "Thank You" From A Kiosk I am a Diving Being : You are an Object You Have No Right To Speak In My Holy Tongue
Social Democrat
the "three dots" menu options I see on a bluesky post
An email sent to all University of Melbourne staff where HTML tags are visible: <p>The University of Melbourne has been advised of a data incident affecting Canvas LMS.</p> <p>We are working closely with the vendor to determine more details.</p> <p>Please visit the <a class="inline_disabled" href="https://www.unimelb.edu.au/cybersecurity/news-and-alert-stories/news/canvas-lms-incident" target="_blank">Canvas LMS incident page</a> for further information.</p>
xkcd://physicists
bias of different methods for left-censored data full: original (uncensored) data cca: complete case analysis, drop censored observations vglm: left-censored Poisson regression vglm_quasi: left-censored quasi-Poisson regression mean_imp: replace censored observations with mean of possible values mi_vglm: multiple imputation using left-censored Poisson regression mi_vglm_quasi: multiple imputation using left-censored quasi-Poisson regression mi_pois_mean: impute using Poisson distribution fit to the mean of the uncensored observations [from Peter Ellis] mi_unif: impute using uniform distribution - like mean_imp but multiple imputation [from Peter Ellis]
Connex-era Melbourne Metro map (mid-2000s), where colours indicate fare zones, not groups of railway lines
Your result: Neither Somehow. German: 33% Autistic: 20% You are, as far as this diagnostic can establish, neither specifically German in your cognitive habits nor particularly autistic in your neurological profile. You are something rarer in the context of people who take quizzes like this: apparently normal about it.
selfie at the summit of Mount Bogong (tallest peak in Victoria)
You scored 28% That puts you in the "not a tech bro" range
Tweet from lauren @NotABigJerk: i'm not falling for that hot take. that's clearly someone with a fetish for getting yelled at. i refuse to participate in that kind of perversion
xkcd 349 "Success"
screenshot of Garmin Connect sleep analysis
Mount St Phillack summit
1999 "AU" Ford Falcon Utility
delta E_OK JND = 0.0065 "Very nice. Comfortably below the 0.02 JND that the colour science people say is the limit. You'd make an excellent paint-chip inspector. Is that a job? It should be."
https://en.wikipedia.org/wiki/Mal_Meninga section "Political career" Meninga briefly campaigned as an Independent for the seat of Molonglo[44] in the Australian Capital Territory Legislative Assembly during the Territory's 2001 general election, informally aligned with the socially conservative, pro-life grouping of Paul Osborne, MLA for Brindabella.[45][46] On Monday 24 September 2001, after being asked why he was standing, he pulled out mid-sentence, stating, "And the thing about that is, I guess, I was a public figure and I was put on the podium where I was just a person out there ... I'm buggered, I'm sorry, I have to resign."[47] In 2017, he said that in the six weeks of preparation he had done leading up to the interview, he had not considered how he would respond if asked "why should people vote for you?".[48]
sunset at Tawonga Huts
one of the Tawonga Huts
selfie near Langford Aqueduct, Falls Creek
camping gear ready to pack into a backpack
oh no
Tweet from Nick "Tomato King" Huber @sweatystartup: You don't understand scale. Use $50 to buy 10 tomato plants. In 6 months you have 250 tomatoes. Plant those into 250 plants. 6 months you have 6,250 plants. Plant them. 6 months you have 156k tomatoes. Plant them. 6 mo you have 3.9MM tomatoes. Sell them for $1 each.
posts from Sky Marchini and Erin Cook making the same point within a minute of each other
me, in front of a lake
the cool S we all drew in primary school
most people rejects his message. they hated jesus because he spoke the truth.
The Terminator. J, Cameron, A Schwarzenegger, L Hamilton, R Patrick. 1998
The Whitlams - Eternal Nightcap
Screenshot of Wikipedia article "Beer in Australia"
me, at the pub, at Mount Hotham
Siegmeyer of Catarina (Dark Souls)
Anneke Groebler presenting a slide explaining estimands, estimators and estimates using a cake analogy
me presenting at the Australian Statistical Conference
ā€œMy therapy session disguised as a keynoteā€
Perth viewed from the Narrows Bridge on a train to Mandurah
a picture of a ceiling with bright LED downlights
oh no
"On the first, and pivotal, level the decision is a binary oneā€”the choice between zero and nonzero as the size. Validity considerations alone are often sufficient to imply that zero is the optimal size" - O. S. Miettinen "Theoretical Epidemiology" (1985)
xkcd "My code's compiling"
BOTH
Muse - Origin of Symmetry
What if Zelda Was a Girl
Winnie the Pooh: data augmentation Winnie the Pooh in a tux: Bayesian inference with informative priors
Dear All, I am delighted to report to you the first results of an experiment that is being conducted in the tea room of this Department. On 29 August, an Oslo Stainless Steel Cutlery Set (48 Pieces: 12 Knifes, 12 Forks, 12 Tea Spoons, 12 Dinner Spoons) was placed in a drawer in the Departmental tea room. A poster asking not to remove the cutlery from the tea room was attached in a visible location. The original research plan was to monitor (in real time) the disappearance of the cutlery from the tea room and then fit a point/counting process model to the observed data. However, as the research project failed to attract any major grants (and so hiring a research assistant to do the work was impossible), it was decided to conduct opportunistic discrete monitoring only. Here is the first result: sixty days in the experiment, the disappearance rates stand as follows: Knives: 8.3% Forks: 100% Tea spoons: 58.3% Dinner spoons: 41.7% Possible conclusions include, but are not limited to, the following: the people who removed the missing cutlery donā€™t understand written English; they donā€™t see forks as a special case of cutlery; forks are needed to do some kind of mathematics; the knives from the set are no good; small spoons are slightly more attractive than the large ones; things left in our tea room tend to disappear even if they are not edible; further research is needed to fully understand the phenomenon. Best, kostya borovkov
Act One Rosencrantz and Guildenstern are betting on coin flips. Rosencrantz, who bets heads each time, wins 92 flips in a row. The extreme unlikeliness of this event according to the laws of probability leads Guildenstern to suggest that they may be "within un-, sub- or supernatural forces". It is revealed that the duo are journeying to court on the orders of the King. Guildenstern theorizes on the nature of reality, focusing on how an event becomes increasingly real as more people witness it.
Tigger
screenshot of Simon Bazelon tweet with a graph of social media following vs candidate performance. the "trend line" is completely flat from 10 to 1M followers, and then slopes steeply downwards for the 4 candidates with >1M followers
Isn't this just mathematical wank? - Tradeoff questions are critical in clinical trials - Different summaries do give different answers - Statisticians need to resist these being answered by mathematical defaults
xkcd 1319 "Automation"
int getRandomNumber() { return 4; // chosen by fair dice roll. // guaranteed to be random. }
oh no (source: webcomicname.com)
xkcd success.png
ā€œIn post-crash interviews, Engineer B told investigators that he would have easily determined the INU to be the source of the problem had he seen the entry in the technical log, but in the event he didnā€™t see it and he didnā€™t ask. Instead, he did something which is known to increase the risk of maintenance errors: he finished a task he didnā€™t start, and in doing so inadvertently added the weight of his expertise to the erroneous decision to attempt the task in the first place.ā€ - https://admiralcloudberg.medium.com/finding-fault-the-crash-of-korean-air-cargo-flight-8509-36a4c1b7b58e
Melbourneā€™s Southbank skyline from a bar on the river