## 04 Jan Predictive Analytics versus Predictive Modeling

My initial thought was that Predictive Analytics refers to an overall field of expertise while Predictive Modeling refers to an activity in which individuals apply potentially relevant mathematical algorithms to data sets in order to learn its structure which can then be applied to new observations to make predictions.

While I did not find the ‘Predictive Analytics’ and ‘Predictive Modeling’ article on Wikipedia very helpful in clarifying the question (for instance, on the one hand, it stated that predictive analytics * encompasses* data mining, and then, on the other hand, it argued that predictive analytics

**data mining), I found a great series of articles about the topic “Predicting the Future” by Alex Guazzelli, VP of Analytics, Zementis, Inc. on IBM developerWorks. In that article, the author addresses the difference between predictive analytics and predictive modeling.**

*is an area of*And indeed, Guazzelli refers to **Predictive Analytics** as a discipline. Predictive analytics attempts to make use of the huge amount and diversity of data generated in an ever increasing pace in our societies in order to make predictions about the future. It thereby complements the reach and power of Descriptive Analytics which helped and still helps decision makers to make informed decisions based on what happened in the past by analyzing and summarizing historical data only.

In contrast, **Predictive Modeling** is the activity of applying “predictive modeling techniques, the mathematical algorithms that make up the core of predictive analytics” according to Guazzelli. In short, the result of applying a particular or ensemble of predictive modeling techniques to a data set is a predictive model. The process or activity can be referred to as predictive modeling. The predictive model can then be applied to new observations in order to make predictions about future outcomes. Both Guazzelli and the ‘Predictive Modeling’ Wikipedia article name a number of available predictive modeling techniques including logistic regression, generalized linear models, random forests, neural networks, support vector machines, or Naive Bayes.

**In summary**, predictive analytics refers to the discipline of learning the generalizable structure of the data that our societies accumulate and use it to make predictions about the future. Predictive modeling is a major part of predictive analytics which makes use of mathematical algorithms to analyze data, identify patterns in it that may otherwise remain hidden to human experts, and learn the generalizable structure of data. The result is a predictive model that can be applied to new observation and events and help to make predictions about future outcomes.

The four-parts series (links below) provide a more comprehensive overview of predictive analytics, predictive modeling and eventually a predictive solution.

**References:**

- Predicting the future, Part 1: What is predictive analytics?
- Predicting the future, Part 2: Predictive modeling techniques
- Predicting the future, Part 3: Create a predictive solution
- Predicting the future, Part 4: Put a predictive solution to work
- Predictive Analytics on Wikipedia
- Predictive Modeling on Wikipedia

## No Comments