F is for Forecast

I recently wrote a children’s book called D is for Data: The ABCs of Data Analytics. This is the sixth in a series of behind-the-scenes, companion articles that will dive a little deeper into each term. We’ll explore the illustration used to define the term, how the word is used in the data world, and other interesting (to me) trivia.

When I think of a forecast, I think of a weather forecast. People have been trying to predict the weather for a long time with varying success. Many of the breakthroughs made in modern weather forecasting are due to data. Quoting from this weather.gov article “Twice a day, every day of the year, weather balloons are released simultaneously from almost 900 locations worldwide!” These balloons are gathering data like wind speed, temperature and humidity that are fed into computer models that forecast the weather.

There are couple of things that make forecasting easier—One is a lot of data. Too little data and you may not be able to see past the outliers. You need a large quantity of data to make a good prediction. Two, you need timely, fresh data. Of course, you need historical data to train your model, but when you’re making an inference (a conclusion based on evidence), then you need timely data. That’s why they release 900 balloons twice per day and constantly gather data from small weather stations disbursed throughout the world. Have you noticed that the predictions get better as the day approaches? That’s because it’s easier to see what’s coming. The same principal is true with computer models, but what the computer sees is data. If it “sees” old data, it can’t “see” it coming.

Forecasting is not just for weather any more. Sales forecasts are very sophisticated now. And sports analytics can predict the success or failure of many coaching decisions. With all of the data that we have available to us today, it’s best to “go with the numbers!”