Question
Stratosphere Technical Consulting
US
Last activity: 30 Aug 2016 10:51 EDT
Predictive Analytics Semantics
Both predictive and adaptive models fall under the concept of predictive analytics. However sometimes in the student guides "predictive model" is used when I think it should be "predictive analytics".
As an example on page 14 in the PDME Student Guide the following statement is made:
The propensity for each customer/product combination is determined by the predictive models in the decision engine
On page 38 in Decision Management Advanced:
Now, in the Determine Propensity Component, we’re going to use Predictive Analytics to determine the propensity of
each of the remaining propositions. ... In order to calculate the propensity, we use an Adaptive Model component that references the actual
predictive model. The referenced predictive model is configured to monitor customer characteristics like
Age, Region and Salary....The Adaptive Model determines the propensity.
When they use the phrase "predictive models" in both quotations - are they not referring to predictive analytical models which is a superset of both adaptive and predictive models. They are not referring to predictive models that utilize historical data ?
Just want to clarify my understanding.... any comments will be appreciated.
Pegasystems Inc.
NL
Bellerophon,
Thanks a lot for your post.
We totally understand the confusion and are revising the content. We will let you know when the new version is on Pega Academy.
Marco.
Pegasystems Inc.
NL
Bellerophon,
Thank you very much for your message. We agree this is a little confusing so I hope to clarify it with the following explanation.
Actually we reffer to both the adaptive models (automatically created when executing a strategy containing an Adaptive Model component) and the predictive models (created with Predictive Analytics Director) as Predictive Models.
In case of the example you mentioned, we have a startegy which contains an Adaptive Model component.
Inside this component we reffer to an Adaptive Model rule as specified on the Adaptive Model tab.
In the output mapping, we explain that the propensity is set by the adaptive model itself.
And once you save and execute such a strategy, the actual adaptive models are created, which are visible on the landing page:
Bellerophon,
Thank you very much for your message. We agree this is a little confusing so I hope to clarify it with the following explanation.
Actually we reffer to both the adaptive models (automatically created when executing a strategy containing an Adaptive Model component) and the predictive models (created with Predictive Analytics Director) as Predictive Models.
In case of the example you mentioned, we have a startegy which contains an Adaptive Model component.
Inside this component we reffer to an Adaptive Model rule as specified on the Adaptive Model tab.
In the output mapping, we explain that the propensity is set by the adaptive model itself.
And once you save and execute such a strategy, the actual adaptive models are created, which are visible on the landing page:
The actual Adaptive Model rule, referenced in the strategy contains the predictors + other configurations:
Regarding the actual description you are giving:
"In order to calculate the propensity, we use an Adaptive Model component that references the actual
predictive model. The referenced predictive model is configured to monitor customer characteristics like Age, Region and Salary....The Adaptive Model determines the propensity."
We agree it is too confusing. We are changing this to:
In order to calculate the propensity, the Adaptive Model component references the Adaptive Model rule. The referenced Adaptive Model rule is configured to monitor customer characteristics such as Age, Region and Salary.
Does this help?