Do you have data for the past 12 months rather than only October? That way, we can observe how gradual (or not) the curve was.
All data in CSV format would be ideal if someone has it to share via gist or such.
All data in CSV format would be ideal if someone has it to share via gist or such.
EDIT: I generated a new sheet for the chart with up-to-date data and fixed a data issue noted below.
The spike was in March which was the start of the COVID-19 pandemic in the United States/shelter-in-place, which was a time of odd behavior and so a spike wouldn't be too weird.
It turns out that dang posted a special whoishiring soon after which was massively popular: https://news.ycombinator.com/item?id=22665398
I thought I filtered out nonstandard threads in the query:
WITH whoishiring_threads AS (
SELECT id FROM `bigquery-public-data.hacker_news.full`
WHERE `by` = "whoishiring"
AND REGEXP_CONTAINS(title, "Ask HN: Who is hiring?")
)
Data science is fun like that, and surprisingly not the first time I've made that particular query mistake.
More pernicious than ? is . though. Not that it matters in your case, but a lot of matches really can be "oops, a one character substitution totally matches, too".
Although in my work I tend to use REGEXP_CONTAINS() as an efficient multifilter for different inputs, which speeds things up too.
There was a combination of large companies needing programmers to enable remote work or process changes due to the pandemic, stimulus funds causing a lot of VC funding to hit startups, hiring binges at FAANG companies, and everyone suddenly being remote-friendly.