Jonas Oesch

@jonasoesch.ch

Visual journalist at @visuals.nzz.ch Personal: jonasoesch.ch

How to properly AI-generate alt-texts for charts A simple screenshot prompt produces texts that are too long and unstructured. By interviewing blind readers, Alessia Vannini and Julia Locher found a form that works much better. Report and examples: github.com/a-vannini/Al...

A stacked bar chart and the corresponding structured alt-text:

Horizontal bar chart. Title: Career choice follows old role patterns. Subtitle: Shares of women and men in various occupations in 2022, in percent. Horizontal axis: 0 to 100 percent. Vertical axis: Nursing and midwifery (vocational education and training), Electricity and energy (vocational education and training), Social work (students at universities of applied sciences), Engineering and IT (students at universities of applied sciences), Human medicine (students at traditional universities), Mechanical and electrical engineering (students at traditional universities). One bar per occupation, divided into two segments: Women (purple), Men (green).

In most occupational fields, a gender-specific distribution is evident.

🏠 Nice & clean → relatively low rent? 🏡 Gorgeous & beautiful → really expensive? Which adjectives in Craigslist rental posts for SF Bay Area go with which prices? #TidyTuesday | 2022 week 27

Nice and clean - relatively low rent?
Adjectives used to describe houses and apartments
in the San Francisco Bay Area in the titles of rental posts on Craigslist and how they are related to rental prices. Titles from 198,279 rental posts on Craigslist between 2000 and 2018. The 15 most frequent adjectives are shown.
Adjectives associated with prices below grand median: nice, clean, special.
Adjectives associated with prices above grand median: beautiful, single, gorgeous, furnished.

Swiss supermarket chain Migros promised price cuts—and delivered on many. But at the same time, products like chocolate and coffee became more expensive, reflecting global market trends. For most shoppers, the result is only modest savings. www.nzz.ch/ld.1869677

Beeswarm plot showing the change of product prices. A normal distribution with the center around zero.

Hello dataviz and design people! We're looking for a freelancer to help us with data visualization and/or diagrams at Works in Progress. We cover topics like housing, technology, health, energy, and much more, and experience in these areas is a plus! If this is you, get in touch! (details below)

Marius Irgens on distortions in vis: "A more common way to distort in […] visualization is through non-linear color mapping." True. Designing a truthful color mapping is incredibly difficult – if not impossible. But as he notes, you need to be transparent about it. nightingaledvs.com/distortion-b...

Distortion by Design - Nightingale

Explore the role of distortion in data visualization, its ethical implications, and use in enhancing comprehension or misleading audiences.

nightingaledvs.com

NEW: 2024 has just been confirmed as the warmest year on record, and the first to breach the 1.5C threshold. We used a ridgeline (Joy Division inspired) chart to visualise daily temperature anomalies since 1940. 2024 clearly stands out with 100% of its days above 1.3C and 75% above 1.5C.

Ridgeline chart showing the distribution of global daily air temperature differences from the pre-industrial reference period (1850-1900), for every year between 1940 and 2024. Each individual year resembles a hill, shaded in a darker shade of red and further to the right for warmer years. The trend is clearly towards warmer years, with 2024 standing out as first year above 1.5C.

Very important thread. Discontinuing fact checking *is* biased because it vastly favours one side of politics. (The side that is spreading disinformation more often, and if you don't know which side that is, you can find out by looking at the data below. Or any other data, there is a plenty).

David Rand@dgrand.bsky.social · 2y ago

🚨In Nature🚨 Meta is dropping fact-checking to avoid anti-conservative bias- but is there actually evidence of bias? We this test empirically & find that conservatives * ARE suspended more * BUT share more misinfo So suspension isn't necessarily evidence of bias www.nature.com/articles/s41...