Mathew Kiang
@mathewkiang.com
switch(
sample(x = 1:3, n = 1, prob = c(.05, .1, .85)),
tweet_joke(),
tweet_plot(),
dont_tweet()
)
sample(x = 1:3, n = 1, prob = c(.05, .1, .85)),
tweet_joke(),
tweet_plot(),
dont_tweet()
)
I use Slack for precisely one thing.
September 24, 2025 at 5:27 PM
I use Slack for precisely one thing.
I'm just saying I've never took the day off work, went for a bike ride, enjoyed the summer inversion and then regretted it.
August 8, 2025 at 8:31 PM
I'm just saying I've never took the day off work, went for a bike ride, enjoyed the summer inversion and then regretted it.
Making travel decisions at this point in my career:
July 31, 2025 at 4:33 AM
Making travel decisions at this point in my career:
The people change but the coffee stays the same.
June 10, 2025 at 6:21 PM
The people change but the coffee stays the same.
When your child is reading a pick-your-own-adventure book and asks how many different ways you can get to the various endings, and you've got too much time on your hands...
June 3, 2025 at 4:27 AM
When your child is reading a pick-your-own-adventure book and asks how many different ways you can get to the various endings, and you've got too much time on your hands...
Key result 2: We won't see the full impact of policies being made today for years. Our model suggests measles would become commonplace in about 20 years — sooner with lower vaccination rates.
As vaccines are attacked, they'll say, "See? Told you it'll be fine." Reader: it won't be fine.
As vaccines are attacked, they'll say, "See? Told you it'll be fine." Reader: it won't be fine.
April 24, 2025 at 4:19 PM
Key result 2: We won't see the full impact of policies being made today for years. Our model suggests measles would become commonplace in about 20 years — sooner with lower vaccination rates.
As vaccines are attacked, they'll say, "See? Told you it'll be fine." Reader: it won't be fine.
As vaccines are attacked, they'll say, "See? Told you it'll be fine." Reader: it won't be fine.
Key result 1: We are right on the precipice of disaster for measles. Just a small drop (e.g., 10%) would result in millions of cases over 25 years and return to a world where measles is commonplace.
But there's good news! Just a 5% increase would bring us back above the critical threshold!
But there's good news! Just a 5% increase would bring us back above the critical threshold!
April 24, 2025 at 4:19 PM
Key result 1: We are right on the precipice of disaster for measles. Just a small drop (e.g., 10%) would result in millions of cases over 25 years and return to a world where measles is commonplace.
But there's good news! Just a 5% increase would bring us back above the critical threshold!
But there's good news! Just a 5% increase would bring us back above the critical threshold!
The two types of academics at conferences these days:
April 24, 2025 at 1:26 AM
The two types of academics at conferences these days:
I got bit by the weirdest R bug I've seen yet... Turns out seq() and c() don't produce the same representation of 0.85?
![## seq() produces unexpected floats?
df <- data.frame(
a = 1:11,
b = seq(.5, 1, .05),
c = c(.5, .55, .6, .65, .7, .75, .8, .85, .9, .95, 1)
)
df[df$b == .85, ] ## doesn't work
df[df$c == .85, ] ## does work
df$b == df$c ## only FALSE for .85](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:jyf67vqhj6ivlzlkvxqnsuyy/bafkreib5sxvspmbahtlwp3w5mwffdb3ojvsmsijf4mcxzxm2gvpvpqhoqm@jpeg)
April 3, 2025 at 9:00 PM
I got bit by the weirdest R bug I've seen yet... Turns out seq() and c() don't produce the same representation of 0.85?
When your coauthor asks if you think you'll hear back about a manuscript under review today and you've got too much time on your hands...
("Hopefully we won't hear back until at least Thursday...")
("Hopefully we won't hear back until at least Thursday...")
March 18, 2025 at 9:10 PM
When your coauthor asks if you think you'll hear back about a manuscript under review today and you've got too much time on your hands...
("Hopefully we won't hear back until at least Thursday...")
("Hopefully we won't hear back until at least Thursday...")
Me, watching the Census, NIH, CDC, and FDA get dismantled while trying to make rent.
March 4, 2025 at 8:46 PM
Me, watching the Census, NIH, CDC, and FDA get dismantled while trying to make rent.
They capped the NIH indirect cost.
This isn't a surprise — they very literally said they would do this last year. That our institutions didn't have a plan and seem unequipped to fight back is... disappointing.
This isn't a surprise — they very literally said they would do this last year. That our institutions didn't have a plan and seem unequipped to fight back is... disappointing.
February 8, 2025 at 12:42 AM
They capped the NIH indirect cost.
This isn't a surprise — they very literally said they would do this last year. That our institutions didn't have a plan and seem unequipped to fight back is... disappointing.
This isn't a surprise — they very literally said they would do this last year. That our institutions didn't have a plan and seem unequipped to fight back is... disappointing.
Maturing is going from using torrents to download movies and music to using torrents to download CDC data.
February 2, 2025 at 4:25 AM
Maturing is going from using torrents to download movies and music to using torrents to download CDC data.
Every ECR text group right now...
January 24, 2025 at 7:29 PM
Every ECR text group right now...
I mean... this is how I cope.
January 23, 2025 at 3:36 PM
I mean... this is how I cope.
There are some prolific traffic violators out there. One vehicle owes nearly a quarter million (!!) in fines, spread across 2800 citations. All prolific violators get their citations (mostly) in downtown.
In 2024, the top 0.1% (~500 cars) accounted for 3% of *all tickets*. 50% belong to just 15%.
In 2024, the top 0.1% (~500 cars) accounted for 3% of *all tickets*. 50% belong to just 15%.
January 7, 2025 at 4:32 AM
There are some prolific traffic violators out there. One vehicle owes nearly a quarter million (!!) in fines, spread across 2800 citations. All prolific violators get their citations (mostly) in downtown.
In 2024, the top 0.1% (~500 cars) accounted for 3% of *all tickets*. 50% belong to just 15%.
In 2024, the top 0.1% (~500 cars) accounted for 3% of *all tickets*. 50% belong to just 15%.
For example, most citations happen around noon on weekdays... But notice the bumps *within* each hour? Turns out the parking officers heap to the nearest 0 or 5 minute mark. Also, oddly, more parking tickets happen earlier in the hour (?).
January 7, 2025 at 4:29 AM
For example, most citations happen around noon on weekdays... But notice the bumps *within* each hour? Turns out the parking officers heap to the nearest 0 or 5 minute mark. Also, oddly, more parking tickets happen earlier in the hour (?).
Almost got run over the other day. He had a memorable license plate so I looked it up when I got home and found out he had over 20 parking violations in 2024 alone! This, obviously, sent me down a bit of a rabbit hole.
For example, did you know SF issues ~5000 citations a day?
For example, did you know SF issues ~5000 citations a day?
January 7, 2025 at 4:26 AM
Almost got run over the other day. He had a memorable license plate so I looked it up when I got home and found out he had over 20 parking violations in 2024 alone! This, obviously, sent me down a bit of a rabbit hole.
For example, did you know SF issues ~5000 citations a day?
For example, did you know SF issues ~5000 citations a day?
I was inspired by @lisahornung.bsky.social to try to generate my own "Year in Review" using the Strava API, R, and ggplot.
Needs some work and a lot of polish but I think the bulk of the code is there.
Needs some work and a lot of polish but I think the bulk of the code is there.
December 12, 2024 at 8:36 PM
I was inspired by @lisahornung.bsky.social to try to generate my own "Year in Review" using the Strava API, R, and ggplot.
Needs some work and a lot of polish but I think the bulk of the code is there.
Needs some work and a lot of polish but I think the bulk of the code is there.
US Census projection model is just like "shrug I dunno -- just LOCF it... lol"
November 28, 2024 at 3:10 AM
US Census projection model is just like "shrug I dunno -- just LOCF it... lol"
November 25, 2024 at 2:37 AM
Parental deaths due to drugs have been increasing steadily over the whole time period — reflecting the larger overdose crisis in the US. In contrast, parental deaths due to firearms have been fairly stable with a sustained increased risk among Black youth that accelerated in ~2015.
May 4, 2024 at 8:18 PM
Parental deaths due to drugs have been increasing steadily over the whole time period — reflecting the larger overdose crisis in the US. In contrast, parental deaths due to firearms have been fairly stable with a sustained increased risk among Black youth that accelerated in ~2015.
In a new paper lead by Ben Schlüter, we estimated about *1.2 million* youth (<18y) in the US have lost a parent due to drugs or guns since 1999. In 2020, drugs and guns accounted for about one in four parental deaths — about double the proportion in 1999.
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
May 4, 2024 at 8:16 PM
In a new paper lead by Ben Schlüter, we estimated about *1.2 million* youth (<18y) in the US have lost a parent due to drugs or guns since 1999. In 2020, drugs and guns accounted for about one in four parental deaths — about double the proportion in 1999.
jamanetwork.com/journals/jam...
jamanetwork.com/journals/jam...
The Festive 500 is a tradition where people try to bike 500km in the 8 days between Christmas Day and New Year’s Eve. To make it a little extra, I wanted to see if I could do all 500km on the GGP Polo Fields bike track (.68 miles per lap). Here are all 456 laps and the average speed per lap.
February 16, 2024 at 10:37 PM
The Festive 500 is a tradition where people try to bike 500km in the 8 days between Christmas Day and New Year’s Eve. To make it a little extra, I wanted to see if I could do all 500km on the GGP Polo Fields bike track (.68 miles per lap). Here are all 456 laps and the average speed per lap.