ConditionalDistribution: Combine continuous distributions

In terminological/classifier-result:

Description

Combine continuous distributions

Combine continuous distributions

Details

Get a wrapper for multiple (independent) continuous distributions conditioned on a discrete variable

Public fields

classes

the class names

weights

the relative weights of each distribution

dists

the distribution as Distribution objects

Methods

Public methods

• ConditionalDistribution$withDistribution()

• ConditionalDistribution$withRandomDistributions()

• ConditionalDistribution$sample()

• ConditionalDistribution$sampleByClass()

• ConditionalDistribution$getPdf()

• ConditionalDistribution$getInverseCdf()

• ConditionalDistribution$plot()

• ConditionalDistribution$theoreticalMI()

• ConditionalDistribution$theoreticalMean()

• ConditionalDistribution$theoreticalVariance()

• ConditionalDistribution$clone()

---

Method withDistribution()

adds in a Distribution with a class name and weight

Usage

ConditionalDistribution$withDistribution(
  distribution,
  class = distribution$label(),
  weight = 1
)

Arguments

distribution

the pdf as an R6 Distribution object (Distribution$new(fn, fnParams...))

class

the classname

weight

the relative weight of this class

---

Method withRandomDistributions()

adds a set of random (uniform, normal, lognormal) distributions to the composite distribution with sensible (random) defaults

Usage

ConditionalDistribution$withRandomDistributions(n = 2)

Arguments

n

the number of distributions to add

---

Method sample()

produce a set of samples conforming to this distribution, with random discrete value

Usage

ConditionalDistribution$sample(n = 1000)

Arguments

n

the number of samples

Returns

a data frame of samples (labelled x) associated with classes (labelled "class")

---

Method sampleByClass()

produce a set of samples conforming to this distribution with preset discrete value

Usage

ConditionalDistribution$sampleByClass(classVector)

Arguments

classVector

the number of samples

Returns

a data frame of samples with continuous values (labelled x) associated with discrete classes (labelled "y") and a sample id column (labelled i)

---

Method getPdf()

get the pdf of these distributions as a dataframe

Usage

ConditionalDistribution$getPdf(xmin, xmax, resolution = 1001)

Arguments

xmin

- the minimum of the support

xmax

- the maximum of the support

resolution

- the number of points in the pdf

Returns

a datafram containing all classes

---

Method getInverseCdf()

get the inverse cdf of these distributions as a dataframe

Usage

ConditionalDistribution$getInverseCdf(resolution = 1001)

Arguments

resolution

- the number of points in the inverse cdf

Returns

a dataframe containing regular samples of all classes

---

Method plot()

plot this distributions as pdf and cdf

Usage

ConditionalDistribution$plot(
  xmin = self$theoreticalMean() - 3 * sqrt(self$theoreticalVariance()),
  xmax = self$theoreticalMean() + 3 * sqrt(self$theoreticalVariance())
)

Arguments

xmin

- the minimum of the support

xmax

- the maximum of the support

Returns

a ggassemble plot object

---

Method theoreticalMI()

generate the theoretical mutual information for this set of distributions using numerical integration of the underlying functions

Usage

ConditionalDistribution$theoreticalMI()

Returns

a single value for the mutual information of this function

---

Method theoreticalMean()

generate the theoretical mean

Usage

ConditionalDistribution$theoreticalMean()

Returns

a single value for the mean of this function

---

Method theoreticalVariance()

generate the theoretical variance

Usage

ConditionalDistribution$theoreticalVariance()

Returns

a single value for the variance of this function

---

Method clone()

The objects of this class are cloneable with this method.

Usage

ConditionalDistribution$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

terminological/classifier-result documentation built on March 14, 2020, 8:04 a.m.