Using Data Science to Better Predict Demand: Case from Twitter and the Flu Season

Michael Watson Ph.D Partner
Read Time: 2 minutes apprx.
case studies data science public health social media data

A recent Wall Street Journal article, “A Bonanza in Smarter Flu Tracking,” discusses tracking the outbreak of flu to better predict demand.

The article notes that companies selling flu related products can boost sales by timing the ads and arrival of product to a market at the start of the flu season in that area.  If they are too early, no one cares and if they are too late, people have already purchased competing products.

So, by watching Twitter chatter on flu and flu-like symptoms, companies can better predict demand and take action.  Here is an example quote from the article:

“During the last flu season, Clorox sent 30,000 additional cases of disinfecting wipes to six states that were most affected by the flu, including cities such as Weslaco, Texas, and Cicero, N.Y.

“We believe some stores would have been out of stock if we didn’t intervene,” said David Kellis, who oversees the company’s social-media communications. The company reported double-digit sales gains in the three months to March 2013 from disinfecting products, which saw record shipments.”

This is a great example of a company taking a new data feed (in this case Twitter) and using data science techniques (mining the data to see what it means and how it predicts the flu), and using optimization to figure out what to do (in the case of Clorox ship an extra 30,000 cases).