Module gPyC

Functions implemented in C

Functions

[hide private]

chisqprob(...)
Return the cumulative probability of a statistic according to the Chi-squared distribution.

Float

family_bdeu(...)
Return the family component of the BDeu score for a BN family from counts and prior precision.

Float

family_llh(...)
Return the family component of the marginal log-likelihood for a BN family from counts and prior parameters.

Float.

lgammadiff(...)
Return log gamma differences.

Float.

lgh(...)
Return log H.

Imports: gsl_exception

Function Details

[hide private]

chisqprob(...)

Return the cumulative probability of a statistic according to the Chi-squared distribution.

Parameters:

chi2 (double) - the test statistic
dof (double) - the degrees of freedom of the Chi-squared distribution

Returns:

The cumulative probability (a double between zero and one, inclusive). Requires gPy to be comgsl_cdf_chisq_P(chi2, k)

family_bdeu(...)

Return the family component of the BDeu score for a BN family from counts and prior precision.

Called as family_bdeu(interval,ri,data,precision)

Parameters:

interval (int) - The distance between successive child values in data
ri (int) - The number of values for the child variable of the CPT
data (List of ints/floats) - The data for the CPT (in the standard ordering)
precision (Int/Float) - The prior precision.

Returns: Float

The family component of the BDeu score

family_llh(...)

Return the family component of the marginal log-likelihood for a BN family from counts and prior parameters.

Called as family_llh(interval,ri,data,prior)

Parameters:

interval (int) - The distance between successive child values in data
ri (int) - The number of values for the child variable of the CPT
data (List of ints/floats) - The data for the CPT (in the standard ordering)
prior (List of ints/floats) - The Dirichlet parameters. prior[i] is the Dirichlet parameter for data[i]

Returns: Float

The family component of the marginal log-likelihood

lgammadiff(...)

Return log gamma differences.

Called as lgammadiff(x,y)

Parameters:

x (Float/int) - Some number
y (Float/int) - Some number

Returns: Float.

lgamma(x+y) - lgamma(y)

lgh(...)

Return log H.

Called as lgh(counts,alpha)

Parameters:

counts (List) - list of counts
alpha (Float) - Prior precision divided by size of table

Returns: Float.

lgh function