Although we won't be using Watson Analytics as an online recommendation engine, it does make practical sense for us to use Watson Analytics as the tool to analyze user behavior data and then provide recommendations based upon Watson's ability to identify predictors with the highest prediction strength.
Predictive strength measures how well a predictor variable accurately predicts a target. You use predictive strength to compare the various predictors within your data. Predictive strength is typically presented as a percentage. A predictor with a predictive strength of 100% perfectly predicts a target. A predictor can be a single input or a combination of inputs. We'll see this work later in this chapter.
As an example of user recommendation, everyone should know about the www.amazon.com website, and most are likely to also be aware that when Amazon recommends a product on its site, it is not a making a random, or chance, recommendation.
In an article found in Fortune, the following explains:
In this chapter, we will use a file of logged online user behaviors in an effort to identify opportunities for an upstart online retail store to kick-start a program aiming to provide unique individual online shopping experiences.
The premise of this new store is to use Watson Analytics to evaluate available user behavior data, looking at data (either volunteered by the user or determined automatically by the website) to create a customized user experience for each user each time they visit the store's website, in an effort to increase sales.
Information such as images and/or advertisements that were clicked on, previously completed and abandoned purchases, as well as user demographical statistics, will hopefully be used as sources to guide the creation of focused marketing campaigns, coupon offers, and even the look and feel and navigation style of the website, for each shopper.