3 ms·

In the first graph, the "Lowess Smoothing" line seems to me to not intuitively be a smoothing of the "No Smoothing" line. Is that just me, does it look like a g

In the first graph, the "Lowess Smoothing" line seems to me to not intuitively be a smoothing of the "No Smoothing" line. Is that just me, does it look like a good fit to you? Maybe "local polynomial regression" is not a good way to extrapolate from a low number of widely scattered plot points.

timmg 2y ago
I think a lot of "smoothing" algorithms don't work at the ends of the data. Smoothing usually takes in past and future data. At the end of the graph there is no future data. (Maybe that's not the case for this type of smoothing?) I've always thought the smoothed line should be truncated (so that it doesn't show present-day estimates where data is lacking.)
- partitioned 2y ago
  It depends on the type of signal. If it's non-linear and chaotic, sure. If its periodic, or has sustained trends, this approach is the basis for most engineering and science.
  - timmg 2y ago
    Do you think global temperature is periodic or non-linear and chaotic?
    - partitioned 2y ago
      I think it is periodic and has sustained trends.
crubier 2y ago
The signal is very noisy, but there is much more past data that is not shown on that figure, supporting the trend line. The trend line is not just fitting data visible on the time interval of the figure, but also fitting past data.
marcosdumay 2y ago
It's much worse than that, since the next paragraph completely negates that finding with "the anomaly has been at about the same level for over the past 13 months". And then there's the next graph. A bit confusing because it uses a different baseline, but where you can see the very recent trend becoming horizontal. (And can't really see any early trend because you can only distinguish 2 years.) Honestly, I'm not sure what to take from that article. But after a needless sensationalistic introduction, I don't think I can take it at face value. An edit: here's a much clearer view of the anomaly, from a different source, with yet a different baseline: https://ourworldindata.org/grapher/temperature-anomaly https://ourworldindata.org/grapher/temperature-anomaly Very recent values have a really large error bar that the source on the article shouldn't have. But you can't use one to fill the missing data on the other.

timmg 2y ago

I think a lot of "smoothing" algorithms don't work at the ends of the data. Smoothing usually takes in past and future data. At the end of the graph there is no future data. (Maybe that's not the case for this type of smoothing?) I've always thought the smoothed line should be truncated (so that it doesn't show present-day estimates where data is lacking.)

partitioned 2y ago
It depends on the type of signal. If it's non-linear and chaotic, sure. If its periodic, or has sustained trends, this approach is the basis for most engineering and science.
- timmg 2y ago
  Do you think global temperature is periodic or non-linear and chaotic?
  - partitioned 2y ago
    I think it is periodic and has sustained trends.

crubier 2y ago

The signal is very noisy, but there is much more past data that is not shown on that figure, supporting the trend line. The trend line is not just fitting data visible on the time interval of the figure, but also fitting past data.

marcosdumay 2y ago

It's much worse than that, since the next paragraph completely negates that finding with "the anomaly has been at about the same level for over the past 13 months". And then there's the next graph. A bit confusing because it uses a different baseline, but where you can see the very recent trend becoming horizontal. (And can't really see any early trend because you can only distinguish 2 years.) Honestly, I'm not sure what to take from that article. But after a needless sensationalistic introduction, I don't think I can take it at face value. An edit: here's a much clearer view of the anomaly, from a different source, with yet a different baseline: https://ourworldindata.org/grapher/temperature-anomaly https://ourworldindata.org/grapher/temperature-anomaly Very recent values have a really large error bar that the source on the article shouldn't have. But you can't use one to fill the missing data on the other.