JaneShenGunther/TCGADataTool | Paclet Repository

JaneShenGunther`TCGADataTool`

dynamicallyExploreThreshold

dynamicallyExploreThreshold [probabilities,modelMeasurementsObject,actualOutcomes,classes]
	creates a dynamic showing the location on the ROC curve, the confusion matrix, and model quality metrics at different probability thresholds.

Details and Options



Examples

(1)

Basic Examples

(1)

Get example binomial classifier for TCGA-CESC project data and outcome "Clinical::Patient::vital_status":

In[1]:=

exampleDesignMatrix=

exampleDataTCGA

[{"TCGADataModel","TCGACESCClinicalVitalStatusClassifier"},"DesignMatrix"];outcomeProperty=

exampleDataTCGA

[{"TCGADataModel","TCGACESCClinicalVitalStatusClassifier"},"OutcomeProperty"];exampleModel=

exampleDataTCGA

[{"TCGADataModel","TCGACESCClinicalVitalStatusClassifier"},"Model"]

Out[1]=

Get input data for

In[2]:=

classes=Information[exampleModel,"Classes"]

Out[2]=

{Alive,Dead}

In[3]:=

allProbabilities=exampleModel[exampleDesignMatrix,"Probabilities"];allProbabilities〚;;3〛

Out[3]=

{Alive0.815417,Dead0.184583,Alive0.635113,Dead0.364887,Alive0.84173,Dead0.15827}

In[4]:=

actualOutcomes=exampleDesignMatrix〚All,outcomeProperty〛;actualOutcomes〚;;3〛

Out[4]=

{Alive,Dead,Alive}

In[5]:=

measurements=ClassifierMeasurements[exampleModel,exampleDesignMatrix,MissingValueSynthesis"ModeFinding"]

Out[5]=

Classifier Measurements

Classifier method	RandomForest
Number of test examples	307
Accuracy	(92.5 ± 1.5 ) %
Accuracy baseline	(80.5 ± 2.3 ) %
Geometric mean of probabilities	0.718 ± 0.0098
Mean cross entropy	0.331 ± 0.014
Single evaluation time	35.4 ms/example
Batch evaluation speed	490. examples/s