The cross campaign model (XCM) is a machine learning modelling system designed to take features from openrtb auctions and determine the probability of a good behaviour auction from some predetermined definition of good behaviour.
XCM is currently supported for both python2 and python3.
A model has a name and works with the classifier to return information about the model.
This class uses the XCMStore classes to persist all relevant data to the databases.
The XCMStore class is designed to store in perpetuity the XCM model data and any learned techniques.
The XCM models are continuous and so all historic learnings compound.
XCM is a cross campaign model, meaning that the data from multiple campaigns is combined to produce a model. In this way, features can become either prominent irrespective of campaign or are campaign dependent.
To predict with an XCM model, call the predict method on an XCM class:
>>> from xcm.models.store import XCMStore
>>> xcm_store = XCMStore()
>>> model = xcm_store.list(as_list=True)[0]
>>> data = {} # include stuff here from an XCMRecord
>>> model.predict(data)