I need to add few SR properties as Adaptive Model predictors. Adaptive model definition and the strategy is created in Customer class.
Customer class properties can be easily added as adaptive model predictors by added them under "Predictors" section. But in order to add SR class properties as Predictors, the only option I could find out is to added them as "Parameterized predictors". After doing above when I checked the result in pxDecisionResults dataset, I could see some difference in terms of Customer class and SR class property results (please refer attached).
Appreciate if someone can help in checking if this is the right way of adding SR properties as Predictors and results are correct. If not please suggest how this can be achieved.
There shouldn't be any problem, but I would try to play around with the data types and conversions, e.g. put them into a toInt() or cast them to a decimal data type and set the parameter type to decimal, things like that.
Posted: 1 year ago
Posted: 26 Feb 2020 3:54 EST
Iolanda Da Costa Martins (dacoi)
Principal Process and Information Manager
Hi Jogesh, you describe how you are configuring the models but not how the parameters are set in the strategy. Before starting to investigate whether this is a data type problem, I would first check the input chain to see if the values are passed to the model when it is selected in the decision making process.
Parameterized predictors is solution to use the SR property as ADM predictor.
Regarding predictor type, conversion might come with the risk. You should have the code whereever you are calling adm. otherwise, there will be a run-time error. My suggestion is create a duplicate property (name should be different) with the expected type for ADM