An Introduction to Predictive Modeling

Casey BartoFor marketers just starting to put their feet into the pool of big data, the slew of tactics, techniques, and volume of data can be overwhelming.
So, to get more information on how some of the aspects of big data work and what you need to know, I sat down with Jeff and Sophia from our actionable analytics team for knowledge on how you can get started using your data. Today’s focus: what is predictive modeling and how it is different from traditional analytics.

What is Predictive Modeling?

Jeff: “It’s using data either past behavioral data demographic data, any type of data or information to forecast a future behavior through a statistical algorithm that can be applied to however you’d like to get to a specific business goal. For example, if you want to build a model for who’s most likely to open an email you can look at a variety of data elements to build an algorithm of who’s most likely versus least likely to open email or to forecast a certain response rate or open rate etc. It’s really about looking at past behavior to forecast future behavior.”

“You can [use predictive modeling to] predict just about anything. Being able to do predictive modeling correctly is a very unique skill set so most marketers or most corporations are not going to have folks on staff to build predictive models. We obviously provide those services here. What’s great about Knotice is all this data we collect. It all goes back to what your business goal is and what you’re trying to do. So we could build a model to help you forecast your budgeting, or targeting your audiences, who you want to send emails or display banners to, etc. It starts with what’s the business question, what’s the data that you’ve got to kind of align towards that business question and can we accurately predict it. We just need a data set and if someone’s had that behavior in the past there’s something there and you can try and predict future behaviors.”

What’s the Difference Between Predictive Modeling and Using Traditional Analytics?

Sophia: “The major difference between predictive modeling vs. traditional analytics is that it has a forward looking view. It’s looking at what’s going to happen in the future instead of reporting on what’s already happened in the past because what’s already in the past may not necessarily reflect what’s going to happen in the future. For example, if you have opened an email, if you look at past behavior you think maybe you are likely to open future emails, but just because you have opened one probably means that you are not likely to open a future one because you already did that.”

2 Trackbacks/Pingbacks

  1. […] « An Introduction to Predictive Modeling […]

  2. […] some more insight into how predictive modeling works. If you missed parts 1 and 2, check them out here and here. If you have questions for Jeff and Sophia, let us know below and they’ll be in a future […]

%d bloggers like this: