We've discussed previously developing clusters of customers who have historically shown similar or exact behavioral patterns.

In my below visual I have implemented a cluster analysis and was able to identify 5 unique clusters specifically by purchase behavior.

This does not include the influence of that behavior or association by similar timelines. That would be a different observation.

With my analysis I have identified buying patterns by SKU#. By doing so you can calculate the probability of the purchase of one or more specific products as a prediction model.

You could also predict new product inventions for those who have purchased all of your product line to retain them.

This information would be helpful for recommendation engines that can display highly probable products of interest based on each individual with historical purchasing behavior in alignment with new prospects.

This will also aid with e-mail drip campaigns and the overall method of delivering advertisements based on where users left off during their journey with you.