cats via python

python/pylibvw.cc

@@ -41,6 +41,7 @@ const size_t lCONTEXTUAL_BANDIT = 4;
 const size_t lMAX = 5;
 const size_t lCONDITIONAL_CONTEXTUAL_BANDIT = 6;
 const size_t lSLATES = 7;
+const size_t lCONTINUOUS = 8;
 
 const size_t pSCALAR = 0;
 const size_t pSCALARS = 1;
@@ -51,6 +52,7 @@ const size_t pMULTILABELS = 5;
 const size_t pPROB = 6;
 const size_t pMULTICLASSPROBS = 7;
 const size_t pDECISION_SCORES = 8;
+const size_t pACTION_PDF_VALUE = 9;
 
 void dont_delete_me(void* arg) {}
 
@@ -119,6 +121,8 @@ label_parser* get_label_parser(vw* all, size_t labelType)
       return &CCB::ccb_label_parser;
     case lSLATES:
       return &VW::slates::slates_label_parser;
+    case lCONTINUOUS:
+      return &VW::cb_continuous::the_label_parser;
     default:
       THROW("get_label_parser called on invalid label type");
   }
@@ -148,6 +152,10 @@ size_t my_get_label_type(vw* all)
   {
     return lSLATES;
   }
+  else if (lp->parse_label == VW::cb_continuous::the_label_parser.parse_label)
+  {
+    return lCONTINUOUS;
+  }
   else
   {
     THROW("unsupported label parser used");
@@ -176,6 +184,8 @@ size_t my_get_prediction_type(vw_ptr all)
       return pMULTICLASSPROBS;
     case prediction_type_t::decision_probs:
       return pDECISION_SCORES;
+    case prediction_type_t::action_pdf_value:
+      return pACTION_PDF_VALUE;
     default:
       THROW("unsupported prediction type used");
   }
@@ -546,6 +556,11 @@ py::list ex_get_decision_scores(example_ptr ec)
   return values;
 }
 
+py::tuple ex_get_action_pdf_value(example_ptr ec)
+{
+  return py::make_tuple(ec->pred.pdf_value.action, ec->pred.pdf_value.pdf_value);
+}
+
 py::list ex_get_multilabel_predictions(example_ptr ec)
 {
   py::list values;
@@ -923,6 +938,7 @@ BOOST_PYTHON_MODULE(pylibvw)
       .def_readonly("lConditionalContextualBandit", lCONDITIONAL_CONTEXTUAL_BANDIT,
           "Conditional Contextual bandit label type -- used as input to the example() initializer")
       .def_readonly("lSlates", lSLATES, "Slates label type -- used as input to the example() initializer")
+      .def_readonly("lContinuous", lCONTINUOUS, "Continuous label type -- used as input to the example() initializer")
 
       .def_readonly("pSCALAR", pSCALAR, "Scalar prediction type")
       .def_readonly("pSCALARS", pSCALARS, "Multiple scalar-valued prediction type")
@@ -932,7 +948,8 @@ BOOST_PYTHON_MODULE(pylibvw)
       .def_readonly("pMULTILABELS", pMULTILABELS, "Multilabel prediction type")
       .def_readonly("pPROB", pPROB, "Probability prediction type")
       .def_readonly("pMULTICLASSPROBS", pMULTICLASSPROBS, "Multiclass probabilities prediction type")
-      .def_readonly("pDECISION_SCORES", pDECISION_SCORES, "Decision scores prediction type");
+      .def_readonly("pDECISION_SCORES", pDECISION_SCORES, "Decision scores prediction type")
+      .def_readonly("pACTION_PDF_VALUE", pACTION_PDF_VALUE, "Action pdf value prediction type");
 
   // define the example class
   py::class_<example, example_ptr, boost::noncopyable>("example", py::no_init)
@@ -1001,6 +1018,7 @@ BOOST_PYTHON_MODULE(pylibvw)
       .def("get_scalars", &ex_get_scalars, "Get scalar values from example prediction")
       .def("get_action_scores", &ex_get_action_scores, "Get action scores from example prediction")
       .def("get_decision_scores", &ex_get_decision_scores, "Get decision scores from example prediction")
+      .def("get_action_pdf_value", &ex_get_action_pdf_value, "Get action and pdf value from example prediction")
       .def("get_multilabel_predictions", &ex_get_multilabel_predictions,
           "Get multilabel predictions from example prediction")
       .def("get_costsensitive_prediction", &ex_get_costsensitive_prediction,

python/tests/test_cats.py

@@ -0,0 +1,19 @@
+from vowpalwabbit import pyvw
+
+import pytest
+
+def test_cats():
+    min_value = 10
+    max_value = 20
+
+    vw = pyvw.vw("--cats 4 --min_value " + str(min_value) + " --max_value " + str(max_value) + " --bandwidth 1")
+    vw_example = vw.parse("ca 15:0.657567:6.20426e-05 | f1 f2 f3 f4", pyvw.vw.lContinuous)
+    vw.learn(vw_example)
+    vw.finish_example(vw_example)
+
+    assert vw.get_prediction_type() == vw.pACTION_PDF_VALUE, "prediction_type should be action_pdf_value"
+
+    action, pdf_value = vw.predict("| f1 f2 f3 f4")
+    assert action >= 10
+    assert action <= 20
+    vw.finish()

python/vowpalwabbit/pyvw.py

@@ -98,6 +98,8 @@ def example(self, initStringOrDict=None, labelType=pylibvw.vw.lDefault):
             - 4 : lCONTEXTUAL_BANDIT
             - 5 : lMAX
             - 6 : lCONDITIONAL_CONTEXTUAL_BANDIT
+            - 7 : lSLATES
+            - 8 : lCONTINUOUS
             The integer is used to map the corresponding labelType using the
             above available options
 
@@ -149,6 +151,7 @@ def get_prediction(ec, prediction_type):
         - 6: pPROB
         - 7: pMULTICLASSPROBS
         - 8: pDECISION_SCORES
+        - 9: pACTION_PDF_VALUE
 
     Examples
     --------
@@ -176,6 +179,7 @@ def get_prediction(ec, prediction_type):
         pylibvw.vw.pPROB: ec.get_prob,
         pylibvw.vw.pMULTICLASSPROBS: ec.get_scalars,
         pylibvw.vw.pDECISION_SCORES: ec.get_decision_scores,
+        pylibvw.vw.pACTION_PDF_VALUE: ec.get_action_pdf_value,
     }
     return switch_prediction_type[prediction_type]()
 
@@ -263,6 +267,8 @@ def parse(self, str_ex, labelType=pylibvw.vw.lDefault):
             - 4 : lCONTEXTUAL_BANDIT
             - 5 : lMAX
             - 6 : lCONDITIONAL_CONTEXTUAL_BANDIT
+            - 7 : lSLATES
+            - 8 : lCONTINUOUS
             The integer is used to map the corresponding labelType using the
             above available options
 
@@ -501,6 +507,8 @@ def example(self, stringOrDict=None, labelType=pylibvw.vw.lDefault):
             - 4 : lCONTEXTUAL_BANDIT
             - 5 : lMAX
             - 6 : lCONDITIONAL_CONTEXTUAL_BANDIT
+            - 7 : lSLATES
+            - 8 : lCONTINUOUS
             The integer is used to map the corresponding labelType using the
             above available options
 
@@ -1029,6 +1037,8 @@ def __init__(
             - 4 : lCONTEXTUAL_BANDIT
             - 5 : lMAX
             - 6 : lCONDITIONAL_CONTEXTUAL_BANDIT
+            - 7 : lSLATES
+            - 8 : lCONTINUOUS
             The integer is used to map the corresponding labelType using the
             above available options