Society, in general, is very adept at using weather forecasts and past experiences to make near-term personal or business weather impact decisions.
We have all seen how weather can impact an organization’s operations. Many organizations have plans in place to deal with weather. For instance, schools close when needed and add extra days to the school calendar in response. Many grocery stores staff based on expected weather and customer demand, but in general the overall impact to grocery stores is a time-shift of customer purchases and therefore the impact is lessened somewhat (http://www.federalreserve.gov/pubs/feds/2000/200008/200008pap.pdf). Unfortunately, many other businesses like restaurants are impacted by weather and have no substantial way to recoup the lost business due to events like snowstorms and rain.
What if you could better understand the impact of weather and use the insight to take advantage of opportunities or cut costs? Today’s technology will let data scientists analyze weather and other datasets at levels of granularity that are amazing. The starting point is by looking at historical weather first. Historical weather information can be used in analytical models to help determine patterns and relationships with other datasets (like sales, energy usage, etc.). It is easy to understand how energy usage is affected by weather (especially temperature), but other patterns and correlations can also be discovered. One widely mentioned example of unexpected discovery with weather events and retail is a dramatic increase in strawberry Pop-Tarts at Walmart stores ahead of a hurricane (http://www.adweek.com/adfreak/hurricane-coming-get-pop-tarts-21329).
To do these correlations, a data scientist or analyst will use many different methods to help them understand the relationships and patterns between different datasets. Once these patterns and relationships are better understood, current and/or forecast weather can be incorporated to help predict sales, usage or other impacts. This is how historical weather data can benefit a business. Is this process beyond your capabilities? There are a number of independent data scientists and business analytic experts that can assist smaller organizations who don’t have the staff to perform the analytical functions.
Since weather is local, it is imperative that organizations offering or performing the analytical services use the proper weather data for their analyses. In the past, historical weather data has been only available from weather reporting stations, many of which are located near airports. Weather data from stations 10 to 50 miles away may not be very useful, especially if elevation or coastal influences are common. Technology is such that historical weather data down to the ZIP Code or Latitude/Longitude level is now available making it possible to correlate weather with specific locations. After these correlations are understood, it is then possible to use this intelligence in many ways. At the very least, the knowledge of how last year’s sales or energy usage was impacted by weather can assist in making decisions about what to do this year. By using weather forecasts, the same correlations can be applied and actionable decisions can be made for the near-term. Some long-range forecasts (15-90 days out) are granular enough that they can also be used especially if temperature extremes impact your business.