starvine.uvar.uvmodels.uv_base.UVmodel Class Reference

Univariate parameterized model base class. More...

Inheritance diagram for starvine.uvar.uvmodels.uv_base.UVmodel:

Public Member Functions
def	__init__ (self, paramsString, momtype, bounds, args, kwargs)
def	setupGMM (self, data, w=1.0, nMoM=None)
	General Method of Moments setup. More...
def	fitKDE (self, data, args)
	Fit gaussian kernel density fn to data.
def	setLogPrior (self, logPriorFn)
	Set log of prior distribution . More...
def	setupMCMC (self, nwalkers, params, data, wgts=None, args)
	Initilize the MCMC sampler. More...
def	fitMCMC (self, iters, pos=None)
	Execute MCMC model parameter fitting. More...
def	fitMLE (self, data, params0, weights=None, args, kwargs)
	Generic maximum (log)likelyhood estimate of paramers. More...

Public Attributes
	nParams
	nMomentConditions
	internalGMM
	kde
	logPriorFn
	walker_pos
	nwalkers
	sampler

Private Member Functions
def	_pdf (self, x, args)
	Pure virtual pdf model function. More...
def	_logPosterior (self, params, data, wgts)
	Posterior distribution .
def	_paramCheck (input_f)
	Parameter bounds checker. More...
def	_loglike (self, params, data, wgts)

Detailed Description

Univariate parameterized model base class.

Extends the capabilities of the rv_continuous base class of the scipy.stats module with the GMM capabilites of statmodels.sandbox.regression.gmm (general method of moments fitting).

With the combined capability of scipy and statmodels, it is easy to obtain maximum likelyhood and MoM estimators for univariate model parameters.

Methods inherited from scipy.stats.rv_continuous:

rvs(*args, **kwds)  Random variates of given type.
pdf(x, *args, **kwds)  Probability density function at x of the given RV.
logpdf(x, *args, **kwds)  Log of the probability density function at x of the given RV.
cdf(x, *args, **kwds)  Cumulative distribution function of the given RV.
logcdf(x, *args, **kwds)  Log of the cumulative distribution function at x of the given RV.
sf(x, *args, **kwds)  Survival function (1 - cdf) at x of the given RV.
logsf(x, *args, **kwds)  Log of the survival function of the given RV.
ppf(q, *args, **kwds)  Percent point function (inverse of cdf) at q of the given RV.
isf(q, *args, **kwds)  Inverse survival function (inverse of sf) at q of the given RV.
moment(n, *args, **kwds)  n-th order non-central moment of distribution.
stats(*args, **kwds)  Some statistics of the given RV.
entropy(*args, **kwds)  Differential entropy of the RV.
expect([func, args, loc, scale, lb, ub, ...])  Calculate expected value of a
    function with respect to the distribution.
median(*args, **kwds)  Median of the distribution.
mean(*args, **kwds)  Mean of the distribution.
std(*args, **kwds)  Standard deviation of the distribution.
var(*args, **kwds)  Variance of the distribution.
interval(alpha, *args, **kwds)  Confidence interval with equal areas around the median.
__call__(*args, **kwds)  Freeze the distribution for the given arguments.
fit(data, *args, **kwds)  Return MLEs for shape, location, and scale parameters from data.
fit_loc_scale(data, *args)  Estimate loc and scale parameters from data using 1st and 2nd moments.
nnlf(theta, x)  Return negative loglikelihood function.

Member Function Documentation

_loglike()

def starvine.uvar.uvmodels.uv_base.UVmodel._loglike (   self,   params,   data,   wgts  )

private

Parameters

params	array-like model parameters
data	np_array input data
wgts	np_array input data weights

_paramCheck()

def starvine.uvar.uvmodels.uv_base.UVmodel._paramCheck (   input_f )

private

Parameter bounds checker.

Returns

Typical function value if params are in bounds, -inf if out of bounds

_pdf()

def starvine.uvar.uvmodels.uv_base.UVmodel._pdf (   self,   x,   args  )

private

Pure virtual pdf model function.

Note: the number of paramers is "infered" from the signature of the _pdf OR _cdf methods.

fitMCMC()

def starvine.uvar.uvmodels.uv_base.UVmodel.fitMCMC	(		self,
			iters,
			pos = None
	)

Execute MCMC model parameter fitting.

Must call setupMCMC before calling this method.

Parameters

iters	int number of MCMC iterations
pos	Initial position of walkers

fitMLE()

def starvine.uvar.uvmodels.uv_base.UVmodel.fitMLE	(		self,
			data,
			params0,
			weights = None,
			args,
			kwargs
	)

Generic maximum (log)likelyhood estimate of paramers.

Parameters

data	Input data to fit to
params0	Initial guess for model parameters
weights	Input data weights

Note: compare to scipy's model.fit() method results. Scipy's model.fit() method does not allow for weights :(

setLogPrior()

def starvine.uvar.uvmodels.uv_base.UVmodel.setLogPrior	(		self,
			logPriorFn
	)

Set log of prior distribution

.

Parameters

logPriorFn

function logPrior

setupGMM()

def starvine.uvar.uvmodels.uv_base.UVmodel.setupGMM	(		self,
			data,
			w = 1.0,
			nMoM = None
	)

General Method of Moments setup.

Parameters

data	Input univariate data
w	Data weights

Example General Method of Moments use:

>>> my_uvmodel = UVmodel(...)
>>> params0 = np.array([2.0, 1.0])
>>> my_uvmodel.internalGMM.fit(params0, maxiter=2, optim_method='nm', wargs=dict(centered=False))

setupMCMC()

def starvine.uvar.uvmodels.uv_base.UVmodel.setupMCMC	(		self,
			nwalkers,
			params,
			data,
			wgts = None,
			args
	)

Initilize the MCMC sampler.

Parameters

nwalkers	int Number of walkers
params	list of doubles Model parameter list
data	np_array data array
wgts	np_array data weights

---

The documentation for this class was generated from the following file:

• starvine/uvar/uvmodels/uv_base.py