本文整理汇总了Python中sklearn.decomposition.PCA.components_方法的典型用法代码示例。如果您正苦于以下问题：Python PCA.components_方法的具体用法？Python PCA.components_怎么用？Python PCA.components_使用的例子？那么, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类sklearn.decomposition.PCA的用法示例。
在下文中一共展示了PCA.components_方法的6个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: test_init
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
def test_init(self, df_norm, n_components):
from flotilla.compute.decomposition import DataFramePCA
test_pca = DataFramePCA(df_norm, n_components=n_components)
true_pca = PCA(n_components=n_components)
true_pca.fit(df_norm.values)
pc_names = ['pc_{}'.format(i + 1) for i in
range(true_pca.components_.shape[0])]
true_pca.components_ = pd.DataFrame(true_pca.components_,
index=pc_names,
columns=df_norm.columns)
true_pca.explained_variance_ = pd.Series(
true_pca.explained_variance_, index=pc_names)
true_pca.explained_variance_ratio_ = pd.Series(
true_pca.explained_variance_ratio_, index=pc_names)
true_pca.reduced_space = true_pca.transform(df_norm.values)
true_pca.reduced_space = pd.DataFrame(true_pca.reduced_space,
index=df_norm.index,
columns=pc_names)
npt.assert_array_equal(test_pca.X, df_norm.values)
pdt.assert_frame_equal(test_pca.components_,
true_pca.components_)
pdt.assert_series_equal(test_pca.explained_variance_,
true_pca.explained_variance_)
pdt.assert_series_equal(test_pca.explained_variance_ratio_,
true_pca.explained_variance_ratio_)
pdt.assert_frame_equal(test_pca.reduced_space,
true_pca.reduced_space)示例2: load
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
def load(self, filename='pca.nc'):
"""
Read sklearn PCA parameters from a netcdf file
"""
infile = netCDF4.Dataset(filename, 'r')
self.locations = [json.loads(string) for string in list(infile.variables['location'])]
self.pcas = []
id = 0
for location in self.locations:
n_components = infile.variables['n_components'][id]
components = infile.variables['components'][id]
mean = infile.variables['means'][id]
explained_variance_ratio = infile.variables['explained_variance_ratio'][id]
noise_variance = infile.variables['noise_variance'][id]
pca = PCA(n_components=n_components)
pca.components_ = components
pca.mean_ = mean
pca.explained_variance_ratio_ = explained_variance_ratio
pca.noise_variance_ = noise_variance
self.pcas.append(pca)
id += 1示例3: get_components
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
def get_components(data):
pca = PCA(n_components=COMPONENTS)
pca.fit(data)
new_components = np.array([np.dot(component, ortho_rotation(pca.components_)) for component in pca.components_])
pca.components_ = new_components
print(pca.components_, pca.explained_variance_ratio_)
transformed = pca.transform(data)
df_transformed = pd.DataFrame(data=transformed, index=data.index)
return df_transformed示例4: test_init
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
def test_init(self, df_norm, n_components):
from flotilla.compute.decomposition import DataFramePCA
test_pca = DataFramePCA(df_norm, n_components=n_components)
true_pca = PCA(n_components=n_components)
true_pca.fit(df_norm.values)
pc_names = ['pc_{}'.format(i+1) for i in
range(true_pca.components_.shape[0])]
true_pca.components_ = pd.DataFrame(true_pca.components_,
index=pc_names,
columns=df_norm.columns)
true_pca.explained_variance_ = pd.Series(
true_pca.explained_variance_, index=pc_names)
true_pca.explained_variance_ratio_ = pd.Series(
true_pca.explained_variance_ratio_, index=pc_names)
true_pca.reduced_space = true_pca.transform(df_norm.values)
true_pca.reduced_space = pd.DataFrame(true_pca.reduced_space,
index=df_norm.index,
columns=pc_names)
npt.assert_array_equal(test_pca.X, df_norm.values)
pdt.assert_frame_equal(test_pca.components_,
true_pca.components_)
pdt.assert_series_equal(test_pca.explained_variance_,
true_pca.explained_variance_)
pdt.assert_series_equal(test_pca.explained_variance_ratio_,
true_pca.explained_variance_ratio_)
pdt.assert_frame_equal(test_pca.reduced_space,
true_pca.reduced_space)
# class TestDataFrameNMF():
# def test_init(self, df_nonneg, n_components, RANDOM_STATE):
# from flotilla.compute.decomposition import DataFrameNMF
#
# test_nmf = DataFrameNMF(df_nonneg, n_components=n_components,
# random_state=RANDOM_STATE)
#
# true_nmf = NMF(n_components=n_components, random_state=RANDOM_STATE)
# true_nmf.reduced_space = true_nmf.fit_transform(df_nonneg.values)
# pc_names = ['pc_{}'.format(i + 1) for i in
# range(true_nmf.components_.shape[0])]
# true_nmf.reduced_space = pd.DataFrame(true_nmf.reduced_space,
# index=df_nonneg.index,
# columns=pc_names)
# true_nmf.components_ = pd.DataFrame(true_nmf.components_,
# index=pc_names,
# columns=df_nonneg.columns)
#
# npt.assert_almost_equal(test_nmf.X, df_nonneg.values, decimal=4)
# pdt.assert_frame_equal(test_nmf.components_,
# true_nmf.components_)
# pdt.assert_frame_equal(test_nmf.reduced_space,
# true_nmf.reduced_space)示例5: fit
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
def fit(self, predictors, locations, **kwargs):
self.locations = locations
self.pcas = []
self.n = predictors['n']
for location in locations:
raw = extract_n_by_n(predictors, location, **kwargs)
#pca = PCA(n_components='mle', whiten=True)
#pca = PCA(n_components=0.95, whiten=True)
pca = PCA(n_components=2)
pca = pca.fit(raw)
components = pca.components_
pca.components_ = components
self.pcas.append(pca.fit(raw))
print "pca: ", location, pca.n_components_, pca.explained_variance_ratio_示例6: list
# 需要导入模块: from sklearn.decomposition import PCA [as 别名]
# 或者: from sklearn.decomposition.PCA import components_ [as 别名]
#MAKE DF from BIOM table
col = ['sample']
for i in list(range(8)):
col.append('OTU'+str(i)) #list of column names
table= pd.read_csv('{}/{}'.format(cwd,inputfile),delim_whitespace=True, header= None)
table.columns= col # name columns
table= table.set_index('sample') # index by sample names
null_data = table[table.isnull().any(axis=1)]
print '**Warning: the following lines are missing data. All nulls will be filled with zeros'
print null_data
table= table.fillna(0) #replace any missing values with zeros
table.to_csv(r'pd_df.csv')
print 'pandas dataframe saved as: pd_df.csv'
##REMOVING PC1
pca = PCA(n_components=8) #keeps 8 components? was 100 in original script-not sure why
#X = pca.fit_transform(table.apply(np.log(table))) #fit df into model
X= pca.fit_transform(table)
#print X[1]
Y = X[:,1:] #Y = every value in list X except the first one
#print Y[1] #just showing that the first value is removed
untrans = pca.inverse_transform(X) #get original data matrix back (w/out PC1 variance)
pca.components_ = pca.components_[1:] #remove the first PCA vector
trans = pca.inverse_transform(Y)
print 'new pca vectors saved as: transformed_pca.txt'
with open('transformed_pca.txt' , 'w') as f:
count =0
while count < len(trans):
f.write(trans[count])
count += 1