It’s a White Christmas Somewhere: Mixing Snow and Statistics

The current weather conditions at a trailhead near our headquarters in New Hampshire.

The current weather conditions at a trailhead near our headquarters in New Hampshire.

We’ve all heard the song, White Christmas. Statistically, here in New Hampshire, those dreams usually come true. Recently, Gizmodo posted an article sharing the National Oceanic and Atmospheric Administrations (NOAA)’s predictions of who will be having a white Christmas this year. After looking at the predictions, and looking out my window, I’m pretty skeptical. It got me thinking a lot about what we do here at Rapid Insight on a daily basis.

NOAA is careful to mention their predictions are not based on current meteorological predictions, but based on historical averages from the last 30 years. That highlights a really important element of data-driven decision making. To be certain, NOAA is using data to form these projections, but they’re using a very simple strategy- trend-analysis. In laymen’s-terms, that means they’re looking back at a pattern, and simply extending it forward. That can be risky, because without understanding why the historical events occurred, we don’t know if it’s reasonable to expect going forward or not.

Despite having a white Christmas more than 75% of the last 30 years, to say that Northern New Hampshire has a 75% chance of a white Christmas seems like an extreme exaggeration. Most living in the Northeastern US would agree that this year is a bit of an anomaly. As a result, there’s an error in basing a prediction for a white Christmas on a historical trend that is clearly not holding true this year. Factoring in recent climate conditions, our model would be able to adjust for the warmer temperatures that we are seeing. While trend analyses can be very beneficial, in this case we can see how those predictions fail to paint an accurate picture of today’s environment.

christmas-stats

This is especially relevant to the area that we live and work in, as there are 6 different ski areas within 35 miles of our office. If these ski areas had based the current year around this same trend analysis used in the NOAA report, they would be sadly disappointed. Or possibly, they would have invented a new sport- mud skiing. This isn’t to say that historical trends are meaningless, but instead highlights just how badly they can be wrong when we ignore the many predictors we have available, like temperatures, and unusual weather systems. It doesn’t look like the ski areas will see snow by Christmas and most likely not until the new year.

Here’s hoping for a snowy 2016. Happy Holidays from all of us at Rapid Insight!