Part of my Christmas wish list was a new machine. Your list came at the perfect time. Happy New Year

My book was edited in India while I was writing on Mountain time. We had great daily handoffs.

Never aggregate data, even with millions of records. Ask for sales data and get monthly totals, ask for production data and find "bad days" are not included. Ask for SCADA data and get holes where startup and shutdown are deleted.

This is where the learning happens

This would require mirror image molecules of all food types, vitamins, proteins, carbohydrates, etc. They would starve

Read these >1M numerical records.
One observation has a comma.
"All records are character. I'm sure that's what he means"

Justin:
Good to see you here,
Alastair

Yes, template was a little strong. I think it was a video clip at a conference, but can't find it

Did dplyr serve as the template for pandas in python? (heard from a friend of a friend)

Why does this sound so familiar?

I'll just leave this here

I was surprised that ChatGPT did pretty well with "Produce the R code using ggplot2 to make this graph" [attach screenshot picture]

... or contact @jburnmurdoch.bsky.social directly right here. He does this kind of work for FT and produced this graph

Data and some R scripts for this data are available at github.com/hdigital/par...

I've been on this for over a month. This is simply the best single source of learning I have seen. Humour, rigor, encouragement, clarity, logic ⭐⭐⭐⭐⭐

Watch for the after credit segments on the videos

I found that Anthropic and ChatGPT were pretty good at "Given this set of data(attach csv file), write code using ggplot2 to generate this plot (attach plot)". It had correct axis labels, chart title and even matched the font. I played around with it a bit to tune it up, but it was a good first pass

www.dagitty.net/dags.html in the browser - handy for DAGs

Cool. There used to be an app on LinkedIn that did this