def reconstructMatrix(self, kSingularValues, returnMatrix=False):
(sk, Uk, Vk) = self.decomposeTopK(kSingularValues)
if (returnMatrix == True):
return tsUtils.matrixFromSVD(sk, Uk, Vk)
else:
return (sk, Uk, Vk)
def denoisedTS(self, ind, range=True):
NewColsDenoised = tsUtils.matrixFromSVD(self.sk,
self.Uk,
self.Vk,
probability=self.p).flatten(1)
if range:
assert len(ind) == 2
return NewColsDenoised[ind[0]:ind[1]]
else:
return NewColsDenoised[ind]
def denoisedDFNew(self,
D,
updateMethod='folding-in',
missingValueFill=True):
assert (len(D) % self.N == 0)
p = len(D) / self.N
self.updateSVD(D, updateMethod)
NewColsDenoised = tsUtils.matrixFromSVD(self.sk,
self.Uk,
self.Vk[-p:, :],
probability=self.p)
return NewColsDenoised.flatten(1)
def fit(self, keyToSeriesDF):
# assign data to class variables
self._assignData(keyToSeriesDF, missingValueFill=True)
# now produce a thresholded/de-noised matrix. this will over-write the original data matrix
svdMod = SVD(self.matrix, method='numpy')
(self.sk, self.Uk, self.Vk) = svdMod.reconstructMatrix(self.kSingularValues, returnMatrix=False)
self.matrix = tsUtils.matrixFromSVD(self.sk, self.Uk, self.Vk, probability=self.p)
# set weights
self._computeWeights()