Close Menu
    Daten

    Können Murmeltiere den Frühling gut vorhersagen? [OC]

    Share.
    View 26 Comments

    26 Kommentare

    1. YouTooShallLose on

      Since he’s the most accurate

      From the news: New York City’s resident groundhog, Staten Island Chuck, saw his shadow early Monday — meaning six more weeks of dreaded wintry weather.

      In case any one was wondering

    2. USAFacts on

      It’s Groundhog Day again.

      Figured I’d take a break from posting government spending data and get to the real hard-hitting stuff. This is a very important chart about groundhogs (and a tortoise and a prairie dog statue).

      When I set out, I wrongly assumed Punxsatawney Phil was the only weather-predicting groundhog. It turns out there are a ton of groundhogs doing this, and Phil is actually pretty awful at it.

      [NOAA](https://www.noaa.gov/heritage/stories/grading-groundhogs) looked at Groundhog Day predictions and compared them to actual March temperatures over the past 20 years. Only animals with at least 20 years of recorded predictions made the cut. Here’s how Phil fared:

      The results:

      * Phil gets it right 35% of the time — tied for last among qualified groundhogs
      * He’s worse than several taxidermied groundhogs
      * Worse than a mystery animal presumed to be a groundhog
      * Worse than a prairie dog statue
      * Only Mojave Max, a tortoise, performs worse

      Meanwhile, Staten Island Chuck (formally Charles G. Hogg) clocks in at 85% accuracy.

      Earlier today, both Phil and Chuck saw their shadows. So do with that what you will.

      EDIT: Big news. In the time since I made this chart, NOAA updated their page with 2025 data. And Lander Lil (a literal statue of a prairie dog) has taken over the second place!

      >In a stunning upset, prairie dog statue Lander Lil has overtaken General Beauregard Lee for the number two spot! Several other groundhogs improved their accuracy ratings, but not enough to make big changes in the rankings.

    3. gemeloso on

      The definition of “spring” doesn’t really make sense to me here. How much do temps have to match March to be considered spring?

    4. Veeb on

      Oh man, call me an ignorant European but I thought there was only one official groundhog.

    5. Swoah on

      What’s the split on the current Staten Island Chuck and the one that was brutally murdered by the former mayor in 2014

    6. NadaBurner on

      What do you mean west virginia has literally never seen their beloved mascot and it’s still in the top 5 for accuracy

    8. bonbon367 on

      If you had 19 people flip a coin 19 times and then graphed how often each person flipped “heads” it’d probably look a lot like this.

      Still a cute tradition.

    10. ExGavalonnj on

      There is only one Groundhog and apparently he sucks at his job, get it together Phil.

    11. CrimsonStorm on

      Fun visualization!

      I smell some p-hacking going on here, though 🙂 If you squint, it looks like the distribution of accuracy is pretty normal. I’d enjoy digging into the data more, but one thing I can’t figure out is how a „correct“ prediction is defined. My best guess from ([https://www.ncei.noaa.gov/news/what-will-punxsutawney-phils-six-week-weather-prediction-be](https://www.ncei.noaa.gov/news/what-will-punxsutawney-phils-six-week-weather-prediction-be)) is that „early Spring“ is defined as „March temperatures are above average“, which would mean that it should be an even 50/50 split (ignoring the effects of climate change).

      So then a coin flip would be expected to be correct 50% of the time. Though, of course, „cloudy“ vs „sunny“ isn’t a 50/50 chance either and varies from place to place…

      I’m not a statistician, so that’s as far as I can get without pulling out a textbook, but I’d love to see someone else do more with that.

    12. qwigoqwaga on

      I was curious what the odds of some of these were (I’d expect most to be much closer to 50%!), so based on the binomial distribution assuming 20 years of data, the 35% and 65% ones have about a 13% probability of occurring by chance, so everything in that range is probably just random chance. 25% and 75% are a 2% chance, which is less common. Those 80%+ ones are less than a 1% chance though!

    13. farsightxr20 on

      This looks suspiciously like a normal distribution centered around 50%…

    14. FearMyPony on

      Phil gives a bad name to the surprisingly wide field of rodent-based meteorology.
      Also how does a Taxidermied Groundhog (and a statue) predict the weather? The source article doesn’t really explain that part.

    15. notacanuckskibum on

      Missing Lucy the Lobster, but then this seems to be a USA only list.

    16. Chippings on

      Mojave Max is actually pretty accurate if you just do the opposite of what he says.

    19. om_steadily on

      25% accuracy on a binary problem is actually 75% accuracy, if you just ask the question differently.

    20. LupusDeusMagnus on

      I have no idea what this means, but it sounds like some alien custom

    22. DameKumquat on

      The lifespan of a groundhog is typically 2-3 years, maybe up to 6, in the wild, though up to 12-14 in captivity.

      So are these groundhog clans, or random hogs found in a local area each year, or what?

    24. Its_not_him on

      Just don’t ask Gen Beauregard Lee what his great grandparents were doing in the early 1860s…

    25. baconator81 on

      What is the definition of „spring“ though? Does the average temperature need to exceed a certain threshold to be called „spring“ ?

    Leave A Reply