starvine.mvar.mvd.Mvd Class Reference

Multi-variate data class. More...

Inheritance diagram for starvine.mvar.mvd.Mvd:

Public Member Functions
def	__init__ (self, mvdData=pd.DataFrame(), mvdWeights=pd.DataFrame())
def	setData (self, dataDict, weights=None)
	Collect data from dictionary with {str: np_1darray} {key, value} pairs into a pandas dataFrame.
def	plot (self, kwargs)
	generate pairwise scatter plots
def	setUVD (self, uvdList)
	Collect uni-variate data sets into a multivariate data object. More...
def	computeKDEpdf (self, bandwidth=None)
	Computes mulitvariate kernel density function. More...
def	computeCov (self, weighted=False)
	computes cov matrix More...
def	computePC (self)
	Computes principal components of multivariate data set. More...
def	computePCOP (self, retainFracVar=0.95, reducedDim=None)
	Computes a projection matrix. More...
def	applyPCA (self)
	Provides a reduced order view of the current MVD object. More...
def	plotExplainedVar (self)
	Plots fractional explained varience as a function of number of principal components retained.

Public Attributes
	uvdPool
	mvdData
	mvdDataWeights
	nDims
	mvdKDEpdf
	mvdCov
	eig_pairs
	frac_explained_var
	cum_frac_explained_var
	pcW

Detailed Description

Multi-variate data class.

Performs principal component analysis to reduce the dimensionality of a large data set.

Member Function Documentation

applyPCA()

def starvine.mvar.mvd.Mvd.applyPCA

(

self

)

Provides a reduced order view of the current MVD object.

Returns

(Mvd object, eigenValues_cov, eigenVectors_cov, W) Where W is the transformation matrix.

computeCov()

def starvine.mvar.mvd.Mvd.computeCov	(		self,
			weighted = False
	)

computes cov matrix

Parameters

weighted

bool if True, utilizes mvdDataWeights (vols or areas) as frequency weights, essentially counting samples which represent more "area" or "volume" in the domain more times. True by default.

computeKDEpdf()

def starvine.mvar.mvd.Mvd.computeKDEpdf	(		self,
			bandwidth = None
	)

Computes mulitvariate kernel density function.

A kernel density function is constructed by combining many locally supported "mini PDFs". This is a data smoothing operation.

Parameters

<b>float</b>

(optional) bandwidth. Default is to use the "scott" factor:

computePC()

def starvine.mvar.mvd.Mvd.computePC

(

self

)

Computes principal components of multivariate data set.

Provides a measure of explained varience per principal component, the principal compenent directions and magnitudes.

Returns

list [eigen_pairs, frac explained variance, cummulative_explained_var]

computePCOP()

def starvine.mvar.mvd.Mvd.computePCOP	(		self,
			retainFracVar = 0.95,
			reducedDim = None
	)

Computes a projection matrix.

Maps from the original data space to a reduced param space.

Parameters

retainFracVar	double Desired fraction of explained varience to retain
reducedDim	int Target reduced data dimension

Returns

np_ndarray PC projection matrix

setUVD()

def starvine.mvar.mvd.Mvd.setUVD	(		self,
			uvdList
	)

Collect uni-variate data sets into a multivariate data object.

Parameters

uvdList

list of uvar.Uvd instances

---

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

• starvine/mvar/mvd.py