from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import TruncatedSVD
from scipy.sparse import csr_matrix
from sklearn import datasets
from numpy import count_nonzero
 
# load the inbuild digits dataset
digits = datasets.load_digits()
 
print(digits.data)
 
# shape of the dense matrix
print(digits.data.shape)
 
# standardizing the data points
X = StandardScaler().fit_transform(digits.data)
print(X)
 
# representing in CSR form
X_sparse = csr_matrix(X)
print(X_sparse)
 
# specify the no of output features
tsvd = TruncatedSVD(n_components=10)
 
# apply the truncatedSVD function
X_sparse_tsvd = tsvd.fit(X_sparse).transform(X_sparse)
print(X_sparse_tsvd)
 
# shape of the reduced matrix
print(X_sparse_tsvd.shape)

print("Original number of features:", X.shape[1])
print("Reduced number of features:", X_sparse_tsvd.shape[1])