Measure the Performance of a Predictor

Prepare the predictor function:

In[1]:=

p=Predict[ExampleData[{"MachineLearning","BostonHomes"},"TrainingData"],PerformanceGoal"Quality"]

Out[1]=

PredictorFunction

Input type:

Mixed

(number: 13)

Method: GradientBoostedTrees



Get data on the performance of the predictor on a test set:

In[2]:=

pm=PredictorMeasurements[p,ExampleData[{"MachineLearning","BostonHomes"},"TestData"]]

Out[2]=

Predictor Measurements

Predictor method	GradientBoostedTrees
Number of test examples	168
Standard deviation	4.01 ± 0.56
Standard deviation baseline	8.98 ± 0.71
R-squared	0.801 ± 0.064
Mean cross entropy	4.08 ± 0.72
Single evaluation time	9.61 ms/example
Batch evaluation speed	1.75 examples/ms