Question

Adaptive Decision Manager updates by pxAdaptiveAnalytics

Hi,

I have a question regarding Adaptive model

In my application, if my strategy uses Adaptive Model , then is it needed to store data in pxAdaptiveAnalytics explicitly (by means of data flow or something else)? I am of the impression that pxAdaptiveAnalytics is the data set that is responsible for updating the adaptive models.

To summarise, my question is, in order to use adaptive model , can I simply use adaptive model in strategy or should I also have means to insert data into pxAdaptiveAnalytics data set?

Thanks in advance.

Comments

Keep up to date on this post and subscribe to comments

Pega
July 21, 2019 - 2:38am

Hi,

No, if you directly use adaptive model rule inside a strategy it doesn't store the data in pxAdaptiveAnalytics data set. We use ADm rule in strategy to generate scoring models and later we need to update pxAdaptiveAnalytics dataset with responses.

Following is the example, please check for your reference.

We need to score the adaptive models first and then you need to learn the model, so that it will be accurate while using.

Scoring:

Based upon the ADM rule context different model factories(model instances) will get created.

To do this create an ADM rule first and then include this in the strategy. Now configure a data flow and add this strategy in the data flow. Check the below screen shot.

Strategy:

Let's say I have two propositions in my test group, then two adm model factories will get created. With two different pyname(proposition) values.

Dataflow:

Using this data flow will run the strategy and adm instances will get created. For strategy properties 'Make decision and store data for later response capture'

Learning ADM :

Configure strategy and set pyOutCome property of ADM rule.

In Dataflow use this strategy and write it to pxAdaptiveAnalytics dataset so that ADM will get learned.Give customer data as input to the strategy. In strategy configure choose mode as 'capture response for previous decision in the past period' and give number of days(Delayed learning).

Once you run this dataflow the adm will get learned and store the outcome for the given customer.

Hope it helps.

Pega
July 19, 2019 - 11:56pm

pxAdaptiveAnalytics updates adaptive models linked to a given proposition with the supplied response
This data set is used to train adaptive models.

July 22, 2019 - 8:10am

I also recommend to check out the following course:

https://academy.pega.com/library/72/implementing-adaptive-decisioning

Thanks,

Ivar