We do data science. Smokeless. Proudly based in Madrid. https://nitter.cf/t.co/VDuFFve7Iv
Joined November 2016
- Tweets13
- Following14
- Followers56
- Likes0
Circiter · We do data science retweeted
Recuerdo una charla con el grupo de usuarios de R en Madrid donde les hable de cómo usaba R en el futbol y les comenté un tema que en aquel momento me costaba bastante aceptar.
Les pregunte donde trabajaban: Carrefour, Orange, consultoras varias, etc y les comenté:
'Da igual si hago buen o mal trabajo, podéis entrar ahora a ver como va el Leeds en la clasificación y en 10 segundos podéis juzgar mi trabajo, o yo al menos siento que si la pelota no entra por muy bien que hiciera lo trabajo no era completo."
En cambio, quien sabe (expertos si pero hablo de a pie de calle) si el churn o ARPU de Orange se degrada, si el ticket medio de Carrefour sube o baja, o si esa consulta ha perdido n clientes.
Para mí hice un trabajazo en Leeds, mas 2 años apasionantes, pero nos fuimos a 2a y todo lo q pueda decir o contar será paja ante el kpi mas importante del fútbol, meter mas goles que tus rivales.
Incluso cuento esto con un poco "vergüenza", nos fuimos a 2a...q trabajazo ni que leches. La vida es así. Pero después de un par de años tenía q liberarme de este sentimiento.
E incluso más allá, equipos que ganan ¿quién se pregunta el trabajo que hay detras? La pelota entra y para muchos será suficiente.
De hecho podría contar casos al revés, trabajos que ni fu ni fa, pero envueltos en un éxito q no era mio y en cambio la percepción era diferente de cara afuera.
Circiter · We do data science retweeted
EU GDP flash estimates are in today—the EU grew 1.4% over last year, Eurozone 1.3%
Fastest: Ireland (6.7%, mostly pharma export), Spain (2.6%), Lithuania (2.5%), Czechia (2.4%), Portugal (1.9%)
Slowest: Finland (-.1%), Germany (.4%), Hungary (.5%), Austria (.7%), Italy (.8%)
Circiter · We do data science retweeted
HOW I KNEW EPSTEIN WAS A FRAUD
A mathematician friend of mine was told by Epstein in 2004 that he made his money as a "mathematical option trader". My friend was impressed as Epstein had the largest mansion in Manhattan.
My option friends found no trace of him in the option markets; in the pre-electronic days it was impossible to have a size position w/o being traced (he needed size to make this kind of money). So I knew at 100% there was a scam.
Later was told that he was a "money manager". Again no footprint in the investment world. And supposedly his minimum size was 1 Bn. Impossible.
Epstein tried at several occasions to meet with me through two different people, after he served some jail time. I told them to tell him to fuck off.
So my only contact with him is through his reading of my books.
Circiter · We do data science retweeted
Simulación del crecimiento de la Ciudad de México durante 700 años, entre
1325 y 2025. Se fundó como Tenochtitlan sobre una isla en el Lago de Texcoco, desarrollaron una ingeniería avanzada con chinampas (islas artificiales) para agricultura y diques para separar aguas dulces y saladas. Tras la conquista, el lago fue drenado. Hoy se hunde entre 10 y 30 centímetros por año.
Circiter · We do data science retweeted
Economists occupy an increasing share of cabinets across the world, and it is now the modal degree in cabinets across the world.
This is based on data from Jacob Nyrup and others, and it can be accessed here: ptp.isv.sv.uio.no/ptp/
Unpopular opinion: the Minister of Finance should always be an economist. You need someone with some kind of graduate training in the field (i.e., MSc or PhD) and with some experience outside academia.
If you cannot read research papers and at least understand the gist of them, you're not going to be asking intelligent questions to your staff and you cannot provide sensible policy advice, except perhaps by accident.
Actually, most of your cabinet should be trained in relevant technical fields. Staffing the government with lawyers whose main ability is verbally weasling their way out of explaining their own moronic decisions is wildly suboptimal.
Circiter · We do data science retweeted
Peter Thiel on why all trends are overrated
“I’m always skeptical of sectors and trends. People always ask me what some trends are that I see happening in the future, and I have never liked the question because I’m not a prophet and I don’t think the future is fixed in that sort of way… At this point, I think all trends are overrated… if you hear the words ‘big data’ and ‘cloud computing’, you need to run away as fast as you possibly can. Just think ‘fraud’ and run away.”
He continues:
“All these buzzwords are a tell—like in poker—that the company is bluffing and undifferentiated… We’ve heard the buzzwords before, and so if you’re the nth company in a category that’s well established, that’s problematic… Conversely, I think the things that are underrated are the ones where there are no buzzwords and it doesn’t actually fit into any pre-existing categories… Of course, the challenge is that even the people who are running these companies will describe them in terms of existing categories because that’s so much easier to do.”
One example he uses is that Google would’ve been described as a search engine in 1998. There were already 20+ search engines at the time, but the PageRank algorithm was actually the key differentiating thing. If you simply labeled it as a search engine, that would’ve obscured all of the key differences.
“Figuring out the correct way to think about things in categories for which we don’t even have the proper language is really critical.”
Video source: @twistartups @jason (2015)
Circiter · We do data science retweeted
LLMs play pictionary!
Playing with a new LLM benchmark: how good are they at drawing an SVG of a pelican on a bicycle?
Here's the difference between Claude 3.5 Sonnet new and Claude 3.5 Sonnet previous
ALT Claude 3.5 Sonnet (2024-06-20) - it's clearly a bicycle but the pelican has a pretty small beak
ALT Claude 3.5 Sonnet (2024-10-22) The pelican here has a better sized beak and is also colored an off-white
Muy satisfechos de la felicitación de nuestro cliente por ese pequeño proyecto que usa INLA en el núcleo del núcleo (aunque estamos pensando migrarlo a numpyro en la próxima versión).
Bueno, pues aquí estamos nosotros. elconfidencial.com/tecnologi…
