oskar holm (@ohdatascience) header
oskar holm (@ohdatascience) profile picture

guess what i do.

Joined May 2018
Replying to @erocdrahs
the recent oecd wips conference was all about asking (and not necessarily answering) this question. oecd-events.org/ai-wips-2023… i found this session particularly interesting: oecd-events.org/ai-wips-2023…
1
9
Replying to @r_sanjabi
i follow, and can recommend, @gvwilson.
Replying to @vboykis
i see your point about dashboards, but what is actionable, then?
1
looking forward to this!
our next meeting is this wednesday, the 28th, when @mccarthymorgan, head of data and analytics at @SouthernWater will speak on the importance of data governance for building artificial intelligence services. more info and sign up: meetup.com/brighton-data-for…
1
Replying to @macosken
hi ken! airdrop in aviation is not all bad. did you catch this?nitter.gq/ilove_aviation/status/…
747-8 Captain airdrops pics with Singapore Airlines 777 pilots midair 📸 Loub747
1
oskar holm retweeted
🎉🚨This is real. 🚨🎉 #NormConf is happening. Normconf.com December 15. Online. Free. A day of normcore data takes. It’s gonna be great. Registration is open.
21
181
88
670
Learn how to process & analyse natural language data at tomorrow's @DataBrighton meetup! Owen Eszeki (AI Lead at EDF) will walk us through Natural Language Processing tools for classification & named entity recognition. 🗓 Weds 27th July | 6:15-8:15pm 🎟 meetup.com/brighton-data-for…
1
2
Replying to @pgira
reminds me of the animations that @fluxicon's disco can generate out of log files: fluxicon.com/disco/
1
3
then you learn they are each other's inverse 🤯. it only made sense in the physics class: differentiate wrt to time to move from position(t)-> speed(t)-> acceleration(t). integrate to go back. then, the slope and area representation becomes obvious.
1
5
The .pipe method in Pandas is powerful (yet potentially confusing if you aren't comfortable with passing functions around).
6
19
3
124
come to tomorrow's brighton r meet-up and hear brilliant people discuss their clever problem-solving with #rstats! bonus: i will be presenting, too!
Nothing to do on Thursday evening in Brighton? interested in R? why not join Brighton R for a great evening of discussion meetup.com/Silicon-Brighton-…
Replying to @daniela_witten
this is not meant to be a practical advice, but in case you are curious, you can sign up for a free icloud.com account and run keynote in the cloud for free.
superficially the same. keynote is better at font rendering, object alignment, object selection/edits, arrangements, template edits, image cropping, more (caveat: it has been years since i last used powerpoint, but i used both for many years).
1
1
5
Replying to @daniela_witten
for reasons, i am often forced to use google slides instead of keynote (my preferred option). it cheered me immensely to find someone so aptly describing my frustrations with it: scribe.rip/@laurajavier/goog…
1
9
oskar holm retweeted
Our brand new #infographic has landed! 🥳 From 19,552,563 codes issued 😮 to a 200% increase in employees 👩, it’s been quite a year for Tillo! Check out the full infographic below 👀 ow.ly/omb250HwHox #digitalgiftcards
2
1
No amount of deep learning can fix bad data. Modeling is fun, good data is hard work. Improving data can mean - going through hand-written records - building a labeling interface yourself - find out why data is missing - investigating the "trail" of the data - weeks of curation
12
118
21
773
final reminder: today at 16:00 @justinhowarddev presents "Using AI to measure offline demographics" at the next brighton data forum meeting. sign-up and further info: meetup.com/brighton-data-for…
2
2
Digital Out-of-Home #DOOH Advertising provides a new way to engage with audiences - but how does it target the right people? 🤔 The next @DataBrighton event, will explore how #AI is being used to provide advanced demographic #data for advertising. 🎟️ https://nitter.gq/t.co/XPWX1ZUYoQ
2
2