def __del__(self):
if self.pmml_model:
try:
from pypmml import Model
Model.close()
except:
pass
def is_support(self):
try:
from pypmml import Model
model_content = self.model
if isinstance(model_content, (bytes, bytearray)):
model_content = model_content.decode('utf-8')
if isinstance(model_content, str):
# Check if a file path
if os.path.exists(model_content):
self.pmml_model = Model.fromFile(model_content)
else:
self.pmml_model = Model.fromString(model_content)
return True
else:
Model.close()
return False
except Exception as e:
return False
def test_from_file(self):
# The model is from here: http://dmg.org/pmml/pmml_examples/KNIME_PMML_4.1_Examples/single_iris_dectree.xml
model = Model.fromFile('./resources/models/single_iris_dectree.xml')
self.assertEqual(model.version, '4.1')
app = model.header.application
self.assertEqual(app.name, 'KNIME')
self.assertEqual(app.version, '2.8.0')
self.assertEqual(model.modelElement, 'TreeModel')
self.assertEqual(model.functionName, 'classification')
self.assertEqual(model.modelName, 'DecisionTree')
self.assertEqual(model.algorithmName, None)
self.assertEqual(
model.inputNames,
['sepal_length', 'sepal_width', 'petal_length', 'petal_width'])
inputs = model.inputFields
self.assertEqual(len(inputs), 4)
self.assertEqual(inputs[0].name, 'sepal_length')
self.assertEqual(inputs[0].dataType, 'double')
self.assertEqual(inputs[0].opType, 'continuous')
self.assertEqual(model.targetNames, ['class'])
targets = model.targetFields
self.assertEqual(len(targets), 1)
self.assertEqual(targets[0].name, 'class')
self.assertEqual(targets[0].dataType, 'string')
self.assertEqual(targets[0].opType, 'nominal')
self.assertEqual(model.classes,
['Iris-setosa', 'Iris-versicolor', 'Iris-virginica'])
self.assertEqual(model.outputNames, [
'predicted_class', 'probability', 'probability_Iris-setosa',
'probability_Iris-versicolor', 'probability_Iris-virginica',
'node_id'
])
outputs = model.outputFields
self.assertEqual(outputs[0].feature, 'predictedValue')
self.assertEqual(outputs[1].feature, 'probability')
self.assertEqual(outputs[1].value, None)
self.assertEqual(outputs[2].feature, 'probability')
self.assertEqual(outputs[2].value, 'Iris-setosa')
self.assertEqual(outputs[3].feature, 'probability')
self.assertEqual(outputs[3].value, 'Iris-versicolor')
self.assertEqual(outputs[4].feature, 'probability')
self.assertEqual(outputs[4].value, 'Iris-virginica')
self.assertEqual(outputs[5].feature, 'entityId')
# Data in dict
result = model.predict({
'sepal_length': 5.1,
'sepal_width': 3.5,
'petal_length': 1.4,
'petal_width': 0.2
})
self.assertEqual(result['predicted_class'], 'Iris-setosa')
self.assertEqual(result['probability'], 1.0)
self.assertEqual(result['node_id'], '1')
result = model.predict({
'sepal_length': 7,
'sepal_width': 3.2,
'petal_length': 4.7,
'petal_width': 1.4
})
self.assertEqual(result['predicted_class'], 'Iris-versicolor')
self.assertEqual(result['probability'], 0.9074074074074074)
self.assertEqual(result['node_id'], '3')
# Data in json
result = model.predict(
'[{"sepal_length": 5.1, "sepal_width": 3.5, "petal_length": 1.4, "petal_width": 0.2}]'
)
self.assertEqual(
result,
'[{"node_id":"1","probability_Iris-setosa":1.0,"predicted_class":"Iris-setosa","probability_Iris-virginica":0.0,"probability_Iris-versicolor":0.0,"probability":1.0}]'
)
result = model.predict(
'{"columns": ["sepal_length", "sepal_width", "petal_length", "petal_width"], "data": [[7, 3.2, 4.7, 1.4]]}'
)
self.assertEqual(
result,
'{"columns":["predicted_class","probability","probability_Iris-setosa","probability_Iris-versicolor","probability_Iris-virginica","node_id"],"data":[["Iris-versicolor",0.9074074074074074,0.0,0.9074074074074074,0.09259259259259259,"3"]]}'
)
# Data in list
result = model.predict([5.1, 3.5, 1.4, 0.2])
self.assertEqual(result[0], 'Iris-setosa')
self.assertEqual(result[1], 1.0)
self.assertEqual(result[2], 1.0)
self.assertEqual(result[3], 0.0)
self.assertEqual(result[4], 0.0)
self.assertEqual(result[5], '1')
# Data in list of list
result = model.predict([[5.1, 3.5, 1.4, 0.2], [7, 3.2, 4.7, 1.4]])
self.assertEqual(len(result), 2)
self.assertEqual(result[0][0], 'Iris-setosa')
self.assertEqual(result[0][1], 1.0)
self.assertEqual(result[0][2], 1.0)
self.assertEqual(result[0][3], 0.0)
self.assertEqual(result[0][4], 0.0)
self.assertEqual(result[0][5], '1')
self.assertEqual(result[1][0], 'Iris-versicolor')
self.assertEqual(result[1][1], 0.9074074074074074)
self.assertEqual(result[1][2], 0.0)
self.assertEqual(result[1][3], 0.9074074074074074)
self.assertEqual(result[1][4], 0.09259259259259259)
self.assertEqual(result[1][5], '3')
# Data in numpy
# Shutdown the gateway
Model.close()