I'm looking out the window at a slate-gray sky (again), and thinking about the last weather forecast I saw. That was a couple of days ago, but it called for sunny, warm weather. Not what I'm looking at now. On the other hand, it's the kind of weather that makes me glad to be inside analyzing data and doing statistics.
Since the last weather forecast I got was several days ago, you can probably infer that I'm not one of those people who tune into the Weather Channel obsessively. But one of my co-workers is, and over the summer we decided that it would be fun to track the weather forecasts for a few weeks and see how well the 10-day, 5-day, and next-day forecasts predicted high temperatures.
We wrote an article about it, which you can find on Minitab.com: Weather Forecasts: Just How Reliable Are They?
It was a fun way to apply statistical methods to something that affects all of us every day, and the article illustrates how you can use ANOVA, graphing techniques, hypothesis testing, and regression to draw conclusions from a simple data set.
The article discusses how we collected our data, shows the results of our time series analysis and individual value plots, and runs through what we learned when we used the Assistant in Minitab to help us perform a one-way Analysis of Variance, or ANOVA, then followed it up with some regression analysis.
Based on the data we collected, we concluded that if you're making big plans that could be affected by weather conditions, it's best to rely on the next-day forecast. It makes sense that, given a system as complex as the weather, we'll be better able to predict tomorrow's weather when we know what today's weather is. Or, like Bob Dylan said, "You don't need a weatherman to know which way the wind blows."
If you want to make predictions more than 24 hours out, the five-day forecast is better than using a crystal ball. But the 10-day forecasts exhibit considerably more variation than the other two, so if you're relying on the 10-day forecast for your vacation plans, you might want to pack both shorts and a warm sweater.
Have you collected and analyzed weather data? We'd love to hear about it!