In the Adaptive model of decision strategy, making decision learn from the real-time customer interactions. 

The system learns from customer choices and captures responses, "Next-Best-Action" decision strategy is adaptive in this case.

There are 4 modes for a decision strategy in the adaptive model namely, "Make Decision"(Without storing the customer response), "Make decision and store data for later response capture","Capture response for previous decision by interaction ID"(Considering all the captured responses made in previous decisions for a particular interaction id) and
"Capturing response to the previous decision in the past period"(Considering captured responses made in the previous decision for a particular time period).

In the "Make the decision and store data for later response capture" mode, the inputs used for the prediction/decision are stored in a Cassandra data set. And it's called "pxDecisionResults".

The dataset stores the data into pxdecisionresults using below Keys:

SujectID
pxInteractionID

For additional details please go through the PDN course: IMPLEMENTING ADAPTIVE DECISIONING.

Hope it gives you an idea about "pxDecisionResults". Kindly notify if it does by marking this post as answered/helpful for larger Pega audience.