LearnedDistribution

LearnedDistribution[]

represents a distribution generated by LearnDistribution.

Details and Options

  • The following functions can be used on a LearnedDistribution[]:
  • PDF[dist,]probability or probability density for data
    RandomVariate[dist]random samples generated from the distribution
    SynthesizeMissingValues[dist,]fill in missing values according to the distribution
    RarerProbability[dist,]compute the probability to generate a sample with lower PDF than a given example
  • When acting on a LearnedDistribution[], the functions PDF and RarerProbability can be used with the following options:
  • PerformanceGoalAutomaticaspect of performance to optimize
    MaxIterationsAutomaticnumber of iterations to use when a Monte Carlo integration is performed
    ComputeUncertaintyFalsewhether to return probabilities with their uncertainty
  • Possible settings for PerformanceGoal include:
  • "Quality"maximize the quality of the result
    "Speed"maximize the speed of the result
    Automaticautomatic tradeoff between speed and quality
  • Information[LearnedDistribution[]] generates an information panel about the distribution and its estimated performances.
  • Information[LearnedDistribution[],prop] can be used to obtain specific properties.
  • Information of a LearnedDistribution may include the following properties:
  • "BatchPDFTime"marginal time to appy PDF to one example when a batch is given
    "BatchSamplingTime"marginal time to generate one example in a batch
    "Entropy"estimated entropy of the distribution
    "ExampleNumber"number of training examples
    "FeatureTypes"types of the distribution variables
    "FunctionMemory"memory needed to store the distribution
    "LearningCurve"performance as a function of the training set size
    "MaxTrainingMemory"maximum memory used during training
    "Method"value of Method used by LearnDistribution
    "MethodDescription"summary of the method
    "MethodOption"full method option to be reused in a new training
    "PDFTime"time to apply PDF to a unique example
    "Properties"all information properties available for this distribution
    "SamplingTime"time to sample one example
    "TrainingTime"time used by LearnDistribution to generate the distribution
  • Information properties also include all method suboptions.

Examples

open allclose all

Basic Examples  (3)

Train a LearnedDistribution[] on a numeric dataset:

Look at the distribution Information:

Obtain available information properties:

Generate a new example based on the learned distribution:

Compute the PDF of a new example:

Train a LearnedDistribution[] on a nominal dataset:

Generate a new example based on the learned distribution:

Compute the probability of the examples "A" and "B":

Train a LearnedDistribution[] on a two-dimensional dataset:

Generate a new example based on the learned distribution:

Compute the probability of two examples:

Impute the missing value of an example:

Options  (3)

ComputeUncertainty  (1)

Train a "Multinormal" distribution on a nominal dataset:

A stochastic preprocessing is needed to transform the nominal variables into numeric variables; the PDF computation is approximate:

Use ComputeUncertainty to obtain the uncertainty on the result:

Increase MaxIterations to improve the estimation precision:

MaxIterations  (1)

Train a "Multinormal" distribution on a nominal dataset:

A stochastic preprocessing is needed to transform the nominal variables into numeric variables; the PDF computation is approximate:

Increase MaxIterations to improve the estimation precision:

PerformanceGoal  (1)

Train a "Multinormal" distribution on a nominal dataset:

A stochastic preprocessing is needed to transform the nominal variables into numeric variables; the PDF computation is approximate:

Use PerformanceGoal"Quality" to improve the estimation precision:

Compare with PerformanceGoal"Speed":

Introduced in 2019
 (12.0)