pca.py

# coding:utf-8
importlogging

importnumpyasnp
fromscipy.linalgimportsvd

frommla.baseimportBaseEstimator

np.random.seed(1000)


classPCA(BaseEstimator):
y_required=False

def__init__(self, n_components, solver="svd"):
"""Principal component analysis (PCA) implementation.

 Transforms a dataset of possibly correlated values into n linearly
 uncorrelated components. The components are ordered such that the first
 has the largest possible variance and each following component as the
 largest possible variance given the previous components. This causes
 the early components to contain most of the variability in the dataset.

 Parameters
 ----------
 n_components : int
 solver : str, default 'svd'
 {'svd', 'eigen'}
 """
self.solver=solver
self.n_components=n_components
self.components=None
self.mean=None

deffit(self, X, y=None):
self.mean=np.mean(X, axis=0)
self._decompose(X)

def_decompose(self, X):
# Mean centering
X=X.copy()
X-=self.mean

ifself.solver=="svd":
_, s, Vh=svd(X, full_matrices=True)
elifself.solver=="eigen":
s, Vh=np.linalg.eig(np.cov(X.T))
Vh=Vh.T

s_squared=s**2
variance_ratio=s_squared/s_squared.sum()
logging.info("Explained variance ratio: %s"% (variance_ratio[0: self.n_components]))
self.components=Vh[0: self.n_components]

deftransform(self, X):
X=X.copy()
X-=self.mean
returnnp.dot(X, self.components.T)

def_predict(self, X=None):
returnself.transform(X)