search

jpmml / jpmml-evaluator-spark

PMML evaluator library for the Apache Spark cluster computing system (http://spark.apache.org/)
GNU Affero General Public License v3.0
94 stars 43 forks source link

Spark SQL Analysis Exception cannot resolve column name "Sepal.Length" #3

Closed RamakrishnaAddanki closed 7 years ago

RamakrishnaAddanki commented 7 years ago

Hi folks,

I was trying to deploy the neural network Iris pmml model by using spark and i am facing issue Sparksql analysis exception cannot resolve column name "Sepal.Length" and for more info i am attaching the screenshot of the error can you please help me in figuring out this ......

neuralnetwork_issue

Neural Network iris model i used 

<?xml version="1.0" encoding="UTF-8"?>
<PMML version="4.0" xmlns="http://www.dmg.org/PMML-4_0">
  <Header copyright="morent">
    <Application name="KNIME" version="2.4.1"/>
  </Header>
  <DataDictionary numberOfFields="5">
    <DataField name="sepal_length" optype="continuous" dataType="double">
      <Interval closure="closedClosed" leftMargin="4.3" rightMargin="7.9"/>
    </DataField>
    <DataField name="sepal_width" optype="continuous" dataType="double">
      <Interval closure="closedClosed" leftMargin="2.0" rightMargin="4.4"/>
    </DataField>
    <DataField name="petal_length" optype="continuous" dataType="double">
      <Interval closure="closedClosed" leftMargin="1.0" rightMargin="6.9"/>
    </DataField>
    <DataField name="petal_width" optype="continuous" dataType="double">
      <Interval closure="closedClosed" leftMargin="0.1" rightMargin="2.5"/>
    </DataField>
    <DataField name="class" optype="categorical" dataType="string">
      <Value value="Iris-setosa"/>
      <Value value="Iris-versicolor"/>
      <Value value="Iris-virginica"/>
    </DataField>
  </DataDictionary>
  <NeuralNetwork functionName="classification" algorithmName="RProp" activationFunction="logistic" normalizationMethod="none" width="0.0" numberOfLayers="2">
    <MiningSchema>
      <MiningField name="sepal_length" invalidValueTreatment="asIs"/>
      <MiningField name="sepal_width" invalidValueTreatment="asIs"/>
      <MiningField name="petal_length" invalidValueTreatment="asIs"/>
      <MiningField name="petal_width" invalidValueTreatment="asIs"/>
      <MiningField name="class" invalidValueTreatment="asIs" usageType="predicted"/>
    </MiningSchema>
    <NeuralInputs numberOfInputs="4">
      <NeuralInput id="0,0">
        <DerivedField optype="continuous" dataType="double">
          <FieldRef field="sepal_length"/>
        </DerivedField>
      </NeuralInput>
      <NeuralInput id="0,1">
        <DerivedField optype="continuous" dataType="double">
          <FieldRef field="sepal_width"/>
        </DerivedField>
      </NeuralInput>
      <NeuralInput id="0,2">
        <DerivedField optype="continuous" dataType="double">
          <FieldRef field="petal_length"/>
        </DerivedField>
      </NeuralInput>
      <NeuralInput id="0,3">
        <DerivedField optype="continuous" dataType="double">
          <FieldRef field="petal_width"/>
        </DerivedField>
      </NeuralInput>
    </NeuralInputs>
    <NeuralLayer>
      <Neuron id="1,0" bias="-0.9441846137121959">
        <Con from="0,0" weight="-0.04486308459743835"/>
        <Con from="0,1" weight="-1.007767488246452"/>
        <Con from="0,2" weight="0.4116250900429575"/>
        <Con from="0,3" weight="1.209150565102718"/>
      </Neuron>
      <Neuron id="1,1" bias="-1.5806031030377694">
        <Con from="0,0" weight="0.33364337105537917"/>
        <Con from="0,1" weight="2.454247860268515"/>
        <Con from="0,2" weight="-0.15509338500436265"/>
        <Con from="0,3" weight="-1.8389081867014756"/>
      </Neuron>
      <Neuron id="1,2" bias="-2.7781887628433255">
        <Con from="0,0" weight="0.6896588815822495"/>
        <Con from="0,1" weight="0.4155487213880096"/>
        <Con from="0,2" weight="-0.4693479739569995"/>
        <Con from="0,3" weight="0.024472433148318866"/>
      </Neuron>
      <Neuron id="1,3" bias="-1.8804828048612006">
        <Con from="0,0" weight="-0.046367292285697366"/>
        <Con from="0,1" weight="0.15139622470906453"/>
        <Con from="0,2" weight="0.15126787451340934"/>
        <Con from="0,3" weight="0.5798317894556151"/>
      </Neuron>
      <Neuron id="1,4" bias="-1.0479686251266709">
        <Con from="0,0" weight="0.3022349918305707"/>
        <Con from="0,1" weight="0.5696576400734125"/>
        <Con from="0,2" weight="-1.0984794716289736"/>
        <Con from="0,3" weight="-0.27270023464908766"/>
      </Neuron>
      <Neuron id="1,5" bias="-0.3025179949964037">
        <Con from="0,0" weight="-0.22289067665818585"/>
        <Con from="0,1" weight="-0.5029455541094195"/>
        <Con from="0,2" weight="0.6748013255475972"/>
        <Con from="0,3" weight="0.2769427141691832"/>
      </Neuron>
      <Neuron id="1,6" bias="0.5185804721896162">
        <Con from="0,0" weight="-0.06630369718678882"/>
        <Con from="0,1" weight="0.9366878508504322"/>
        <Con from="0,2" weight="-2.0505238545746756"/>
        <Con from="0,3" weight="-1.4180466765594852"/>
      </Neuron>
      <Neuron id="1,7" bias="-0.14392081477403149">
        <Con from="0,0" weight="0.47767196169403736"/>
        <Con from="0,1" weight="0.9231599358415787"/>
        <Con from="0,2" weight="-1.521293106276155"/>
        <Con from="0,3" weight="-0.9151590899627176"/>
      </Neuron>
      <Neuron id="1,8" bias="-1.1991037918096945">
        <Con from="0,0" weight="-0.0368452580731863"/>
        <Con from="0,1" weight="-0.6438633470456006"/>
        <Con from="0,2" weight="0.09328089052662898"/>
        <Con from="0,3" weight="1.6829695286970527"/>
      </Neuron>
      <Neuron id="1,9" bias="-0.07098521306546789">
        <Con from="0,0" weight="1.1130349585302026"/>
        <Con from="0,1" weight="-0.37623467152890655"/>
        <Con from="0,2" weight="-1.0691145207939827"/>
        <Con from="0,3" weight="-0.6233268000787716"/>
      </Neuron>
    </NeuralLayer>
    <NeuralLayer>
      <Neuron id="2,0" bias="-0.5982891836398166">
        <Con from="1,0" weight="-131.91572569969986"/>
        <Con from="1,1" weight="-2.0803811849944296"/>
        <Con from="1,2" weight="-0.6084039564897556"/>
        <Con from="1,3" weight="-1.066354849840332"/>
        <Con from="1,4" weight="1.7727568679658863"/>
        <Con from="1,5" weight="-8.098634498899418"/>
        <Con from="1,6" weight="19.128469000024445"/>
        <Con from="1,7" weight="47.99808211407523"/>
        <Con from="1,8" weight="-48.02276063336108"/>
        <Con from="1,9" weight="0.8376316854289872"/>
      </Neuron>
      <Neuron id="2,1" bias="0.6887290187921082">
        <Con from="1,0" weight="-16.591716865586278"/>
        <Con from="1,1" weight="0.7787222775775234"/>
        <Con from="1,2" weight="-0.21185970643262134"/>
        <Con from="1,3" weight="-0.5385365682832596"/>
        <Con from="1,4" weight="0.1494344843029574"/>
        <Con from="1,5" weight="-0.2908246414912338"/>
        <Con from="1,6" weight="-209.54696149112175"/>
        <Con from="1,7" weight="41.03251658642203"/>
        <Con from="1,8" weight="-4.481423572323473"/>
        <Con from="1,9" weight="20.57673314463075"/>
      </Neuron>
      <Neuron id="2,2" bias="-0.33370153632947136">
        <Con from="1,0" weight="3.284832540721896"/>
        <Con from="1,1" weight="1.2279556829167277"/>
        <Con from="1,2" weight="0.7602122721765426"/>
        <Con from="1,3" weight="1.1997577238312889"/>
        <Con from="1,4" weight="-11.325671536250567"/>
        <Con from="1,5" weight="1.483961591977191"/>
        <Con from="1,6" weight="-1821.8108555592457"/>
        <Con from="1,7" weight="-148.9767379958393"/>
        <Con from="1,8" weight="6.619929393199292"/>
        <Con from="1,9" weight="-8.681838533973998"/>
      </Neuron>
    </NeuralLayer>
    <NeuralOutputs numberOfOutputs="3">
      <NeuralOutput outputNeuron="2,0">
        <DerivedField optype="categorical" dataType="string">
          <NormDiscrete field="class" value="Iris-setosa"/>
        </DerivedField>
      </NeuralOutput>
      <NeuralOutput outputNeuron="2,1">
        <DerivedField optype="categorical" dataType="string">
          <NormDiscrete field="class" value="Iris-versicolor"/>
        </DerivedField>
      </NeuralOutput>
      <NeuralOutput outputNeuron="2,2">
        <DerivedField optype="categorical" dataType="string">
          <NormDiscrete field="class" value="Iris-virginica"/>
        </DerivedField>
      </NeuralOutput>
    </NeuralOutputs>
  </NeuralNetwork>
</PMML>
vruusmann commented 7 years ago

There is no class org.jpmml.spark.PMMLPredictionModel in the JPMML-Spark codebase anymore. Please upgrade.

RamakrishnaAddanki commented 7 years ago

Hi Thank you very much for your reply I downloaded the latest repo from github just now and tried using same model i am getting same exception for more info see attachment spark_issue1

vruusmann commented 7 years ago

Does your input DataFrame contain sepal_length, sepal_width, petal_length and petal_width columns? The provided NN model needs these four input columns. If they cannot be found, then the model cannot be evaluated.

Your stack trace doesn't seem legit, because it refers to Sepal.Length column (note the capitalization, and the use of the dot character . as a separator instead of the underscore _ character).