A python library for big data analysis. This package is a collection of functions that I have found to be useful in my coursework on big data and machine learning.
This package requires that numpy, scipy, and matplotlib be installed. Using pip this can be done with the following command:
pip install numpy scipy matplotlib
or see Scipy's website for more information on installation.
To install the bigDATA package using pip simply run:
pip install -i https://test.pypi.org/simple/ bigDATA
random
: returns a random matrix of a certain sizeevalues
: returns eigenvalues of a matrixevectors
: returns eigenvectors of a matrixevectors
: returns inverse of a matrixevectors
: returns the covariance matrix of a matrixSVDecomp
: performs a singular value decompositionLUDecomp
: performs an LU decompositionpolarDecomp
: performs a polar decomposition on a given matrix and breaks down the matrix into its rotating and stretching componentssolve
: solves a linear system of equationssolveMany
: Solves for manyx
s in a system of linear equations in the form ofAx=b
where multipleb
's are givenperturb
: perturbs a system of equations and returns the relative perturbation and erroroptimalFit
: given two sets of points, finds the optimal shift and rotation to fit the points in matrixX
ontoY
editDistance
: returns the number of edits needed to turn stringstr1
into stringstr2
jaccardDistance
: returns the percentage of elements in seta
orb
that are not in botha
andb
hammingDistance
: Returns the number ofi
th characters instr1
that don't match thei
th character instr2
cosineDistance
: calculates the cosine distance between lists of numbersa
andb
lrDistance
: calculates the L_r distance betweena
andb
inr
dimensional space
sigmoid
,sigmoid_prime
,softmax
,softmax_prime
, andtansig
activation functionscost
: cost functionClassification
:feed_forward
: feed input forward through a neural networkback_propagation
: performs a single back propagation iteration
KNN
:classify
: k nearest neighbors classification with discrete labelskRegression
: locally weighted kernel regression