Predictive analytics

Predictive analytics is the science concerned with finding repeatable patterns in our data. Such patterns must be reliable enough to serve as a basis for predicting the future behavior of customers and improve business decisions. For predictive analytics algorithms to work we must have access to historical data which exhibits known customer behavior. We must also know what problem we are trying to solve. In the data, we identify fields which are best suited to determine the desired outcome. These fields are called predictors and are combined into a predictive model which you can use in business processes.

Business Value

Let’s position Predictive Analytics in the knowledge landscape and show how it drives the business value.

As you can see, as the business value provided by various tools increases, the complexity of these tools increases. Guesswork is simpler than reporting, which tells us what has happened. In turn reporting is simpler than Analytics, which tells you why it has happened. Analytics is in turn simpler than Monitoring, which reports what is happening now. By adding predictions about what will happen, we further increase the business value but at a cost of a further complexity increase. This meant that in the past the science of predictive analytics was a domain of very few. Only Data Scientists and Mathematics PhDs could use that technology and the tools. Pega Decision Management reduces the complexity of predictive analytics. It empowers data-savvy businesspeople to discover patterns within their data and to develop powerful and reliable models that can predict the customer behavior - the foundation for successful customer interaction decisions everywhere.

Business Application

Predictive models can help in a variety of business dimensions including sales, retention, risk, recommendations, and many others.

A typical application of a predictive model is in a Customer Journey. For example, when a customer calls in to a contact center with a query. The call center agent can process the request based on simple business rules. Alternatively, we can add extra dimensions to the way the call center handles the case by evaluating the current status of the relationship with this specific customer, using a predictive model. We get value from predictive analytics by orchestrating action through decisions, events, and cases empowered by smart analytics at the right time, location, needs, and even mood.

In Pega Decision Management the predictive modelling functionality is delivered through the Prediction Studio portal where you can execute a five-step wizard that guides you through the process of creating a predictive model. The wizard supports a streamlined process for fast model creation. These models can be directly actioned from process flows and decision strategies.

Model creation involves five steps:

Step 1. Data preparation

You identify the decision requiring a predictive model, define the behavior you want to predict, and select the data source. The data is split into development, validation and test sets.

Step 2. Data analysis

You analyze individual fields and determine their role (predictor, ignored, value or benchmark). You can also add virtual fields.

Step 3. Model development

You analyze how the predictors work together and then create predictive models using regression and decision trees.

Step 4. Model analysis

You can compare the models with each other to assess their performance. Typically, you will look at their behavior prediction capabilities and how they segment customers into classes according to predicted behavior.

Step 5. Model selection

You generate the model, including the definition of which fields the model should output. You can also make the model available to strategies and action flows by saving it as a predictive model rule configuration.

Predictive analytics is about linking the way in which cases behave and the information known about them. Prediction Studio links in a mathematical equation referred to as a model. This equation enables you to use the information about a case to calculate a value that predicts the behavior of the case. The specific meaning of the value depends on the type of the model. Pega Decision Management supports two types of models:

• Scoring models - for the prediction of binary behavior, also known as Binary models
• Spectrum models - for the prediction of continuous behavior, also known as Continuous models

Scoring models

The value calculated by the model, known as the score, places a case on a numerical scale. High scores are associated with a high likelihood of the predicted behavior. Typically, the range of scores is broken into intervals of increasing likelihood of one of the two types of behavior (binning). Scoring models require behavior to be classified into two distinct forms like positive and negative. Classic examples of such behavior are:

• Responding to a mailing or not
• Repaying loans or going into arrears
• Making an insurance claim or not

You can also create extended scoring models which can include cases where the behavior is unknown. For example, someone who has been refused a loan cannot repay it or go into arrears.

Spectrum models

Spectrum models extend the ideas of scoring models to the prediction of continuous behavior. The score calculated for each case places it on a spectrum from the lowest value to the highest value. As with scoring models, the score range is broken into intervals. Associated with each interval is the average value of the development sample cases that fall into the interval. These average values provide the predicted value for new cases falling into each interval. Typical applications of Spectrum models include the prediction of:

• Likely purchase value of responders to a direct mail campaign
• The likely eventual write-off of cases recently falling into arrears
• The likely claim value of health, motor or contents insurance policy holders

A spectrum model allows you to differentiate between good, better, and best.

Model templates

When starting to build a new model, you will be presented with the possibility to create a model on a template, that is used for streamlining model development. Templates are available in the risk, retention, recruitment and recommendation categories.

You are of course not limited to these business issues and you can add other templates if required.