Sign in

iiinova is part of the Atmosphere — an open, public network where you own your identity.

Eric Maugendre about data

@maugendre.hachyderm.io.ap.brid.gy
8 followers 2 following 27 posts

Data scientist. Conversations on data responsibility, tech diversity, stats 🌉 bridged from ⁂ hachyderm.io/@maugendre, follow @ap.brid.gy to interact

PostsRepliesMedia
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 07/10/2026
A recent journal article attempts extending biological evolution from #ML! Here is a stylized diagram of "double descent" by the paper author Steve Frank. The first descent (decrease in error on out-of-sample data) occurs as you add input variables to machine […] [Original post on hachyderm.io]
Diagram from Steve Frank (2026) illustrating the double descent pattern of generalization error
Excerpt from https://academic.oup.com/evolut/article/doi/10.1093/evolut/qpag111/8706652
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 05/10/2026
In other words: • An excellent fit makes the data practitioner obtain an inference engine that generalizes badly. • Similary, a living entity that only fits its environment may fail to degrade gracefully when the environment changes. • A social organisation that is optimised is not resilient.
[[A man is talking to a woman]]
Man: I used to think correlation implied causation.
Man: Then I took a statistics class.  Now I don't.
Woman: Sounds like the class helped.
Man: Well, maybe.
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 05/10/2026
Here is where i come from: If you train your machine on a dataset with too few input variables, you cannot capture the behaviour. (Current analogies may be corporations focused on "indicators" and governments focused on GDP.) When learning statistics, we are […] [Original post on hachyderm.io]
a cloud of x and o is given a frontier of a simple curve (Appropriate fitting) or of an intricate curve (Overfitting)
adapted from Kate Crawford @katecrawford.bsky.social https://www.youtube.com/watch?v=fMym_BKWQzk&t=640s
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 26/09/2026
"Directed Acyclic Graphs (DAGs) and Regression for Causal Inference" An easy walk-through by Susan Alber (2022): health.ucdavis.edu/media-resources/…
[DAG] skin cancer is caused by age and sunburn (via moving to Florida)
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 26/09/2026
"Mediators, confounders, colliders – a crash course in causal inference" by Florian Hartig (2019): theoreticalecology.wordpress.com/20… 🧵 #policy #intervention #graphs #causality […] [Original post on hachyderm.io]
Confounding: Smoking is upstream of Exercise and Lung cancer
Mediation: Immune function sits in between Exercise and Lung cancer

from Florian Hartig at https://theoreticalecology.wordpress.com/2019/04/14/mediators-confounders-colliders-a-crash-course-in-causal-inference/
101 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 19/07/2026
We find that cold blobs significantly increase heat wave duration and magnitude downstream over the European continent. Our results highlight the relevance of the subpolar North Atlantic region for European summer conditions, a region that is marked by large […] [Original post on hachyderm.io]
Sea Surface Temperature Anomaly during the second European heatwave this year. A green circle points at an ocean location 7°C below normal.
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 18/07/2026
The weathers, they are a-changing. We can make sense of pattern evolution. I am starting a thread about Europe's new #climate conditions. 🧵 (You may read stories in the "ALT" descriptions.) #trend #heatwave #dataViz #chart #loess #linearRegression #statistics […] [Original post on hachyderm.io]
The chart is temperatures plotted against years (from 1940 to 2026).
It graphs temperatures averaged over four weeks in European France since 1940 [the month Pétain made his first broadcast to the French people].
A curve in red is made from statistical regression (loess method: smoothness determined with cross-validation) to build a statistical central trend.

We may look at the difference between monthly index and trend:
The red point represents the average of mid-June to mid-July 2026. Its high position is not absurd: it shows within the distribution of variability around the trend.
June 1976 and 2003 were outliers, "black swan" like, events: June and mid-June to mid-July 2003 were 3.7 °C away from the trend, when 2026 June is only 1.7°C.
The same kind of extreme period this year would have shown an average at 2°C more.

The calculations and graph were made by Denis Mongin‬ ‪@denismongin.bsky.social‬
The above comments were built from his. Below comments come from me.

Incidently you may note that 2022 was the year the French President said "Mais qui aurait pu prédire ?" and the monthly index exactly is at the trend curve.

Data was mangled by averaging over the month. This has the benefit of showcasing a general enhancement: across the country, day and night.
Data was uploaded from weather stations located in Europe. The method was chosen by the national weather service.
The method purposely excludes weather stations located in France but not in Europe – such as Guyane, a territory neighbour to the Amazon basin.

Florent Malouda was born in Guyane. He is one of the greatest wingers in the history of the Chelsea football club. Malouda represented the France national football team on 80 occasions from 2004 to 2012, including in two FIFA World Cups and two European Championships. He was part of the French team which reached the 2006 World Cup final.

Among the players selected for the 2026 FIFA World Cup is Mike Maignan. He plays as the goalkeeper for and captains Serie A club AC Milan. Mike Maignan was born in French Guiana, to a Haitian mother and a Guadeloupean father.
124 4
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 14/07/2026
During the second heatwave of 2026, temperature in France was in the band of trend variability. (You may read stories in the "ALT" description.) Hat tip to @denismongin.bsky.social‬ #football #chart #dataViz #loess #linearRegression #statistics #climate […] [Original post on hachyderm.io]
The chart is temperatures plotted against years (from 1940 to 2026).
It graphs temperatures averaged over four weeks in European France since 1940 [the month Pétain made his first broadcast to the French people].
A curve in red is made from statistical regression (loess method: smoothness determined with cross-validation) to build a statistical central trend.

We may look at the difference between monthly index and trend:
The average of mid-June to mid-July 2026 is not absurd. It is shown within the distribution of variability around the trend.
June 1976 and 2003 were outliers, "black swan" like, events: June and mid-June to mid-July 2003 were 3.7 °C away from the trend, when 2026 June is only 1.7°C.
The same kind of extreme period this year would have shown an average at 2°C more.

The calculations and graph were made by Denis Mongin‬ ‪@denismongin.bsky.social‬
The above comments were built from his. Below comments come from me.

Incidently you may note that 2022 was the year the French President said "Mais qui aurait pu prédire ?" and the monthly index exactly is at the trend centre.

Data was mangled by averaging over the month. This has the benefit of showcasing a general enhancement: across the country, day and night.
Data was uploaded from weather stations located in Europe. The method was chosen by the national weather service.
The method purposely excludes weather stations located in France but not in Europe – such as Guyane, a territory neighbour to the Amazon basin.

Florent Malouda was born in Guyane. He is one of the greatest wingers in the history of the Chelsea football club. Malouda represented the France national football team on 80 occasions from 2004 to 2012, including in two FIFA World Cups and two European Championships. He was part of the French team which reached the 2006 World Cup final.

Among the players selected for the 2026 FIFA World Cup is Mike Maignan. He plays as the goalkeeper for and captains Serie A club AC Milan. Mike Maignan was born in French Guiana, to a Haitian mother and a Guadeloupean father.
026 6
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 30/03/2026
World total population, in years 1–2100 (log scale), from Danny Dorling's book Slowdown #demographics #charts #dataViz
World total population, in years 1–2100 (log scale), from Danny Dorling's book Slowdown.

We can see big losses in 1500–1600, 1820–1850 and 2020–2100.
100 1
Eric Maugendre about data @maugendre.hachyderm.io.ap.brid.gy · 23/05/2025
AI does not reason. It applies probabilities. It’s like eating an apple and assuming that everything will be as tasty: data.yt/kit/how-ai-is-made.html
A dog is shown an apple core with seeds
124 4