@@ -689,6 +689,55 @@ def _simple_new(
689689 new ._dtype = dtype
690690 return new
691691
692+ @classmethod
693+ def from_spmatrix (cls , data ):
694+ """
695+ Create a SparseArray from a scipy.sparse matrix.
696+
697+ .. versionadded:: 0.25.0
698+
699+ Parameters
700+ ----------
701+ data : scipy.sparse.sp_matrix
702+ This should be a SciPy sparse matrix where the size
703+ of the second dimension is 1. In other words, a
704+ sparse matrix with a single column.
705+
706+ Returns
707+ -------
708+ SparseArray
709+
710+ Examples
711+ --------
712+ >>> import scipy.sparse
713+ >>> mat = scipy.sparse.coo_matrix((4, 1))
714+ >>> pd.SparseArray.from_spmatrix(mat)
715+ [0.0, 0.0, 0.0, 0.0]
716+ Fill: 0.0
717+ IntIndex
718+ Indices: array([], dtype=int32)
719+ """
720+ length , ncol = data .shape
721+
722+ if ncol != 1 :
723+ raise ValueError (
724+ "'data' must have a single column, not '{}'" .format (ncol )
725+ )
726+
727+ # our sparse index classes require that the positions be strictly
728+ # increasing. So we need to sort loc, and arr accordingly.
729+ arr = data .data
730+ idx , _ = data .nonzero ()
731+ loc = np .argsort (idx )
732+ arr = arr .take (loc )
733+ idx .sort ()
734+
735+ zero = np .array (0 , dtype = arr .dtype ).item ()
736+ dtype = SparseDtype (arr .dtype , zero )
737+ index = IntIndex (length , idx )
738+
739+ return cls ._simple_new (arr , index , dtype )
740+
692741 def __array__ (self , dtype = None , copy = True ):
693742 fill_value = self .fill_value
694743
@@ -1900,27 +1949,32 @@ def _make_index(length, indices, kind):
19001949# ----------------------------------------------------------------------------
19011950# Accessor
19021951
1952+
1953+ class BaseAccessor (object ):
1954+ _validation_msg = "Can only use the '.sparse' accessor with Sparse data."
1955+
1956+ def __init__ (self , data = None ):
1957+ self ._parent = data
1958+ self ._validate (data )
1959+
1960+ def _validate (self , data ):
1961+ raise NotImplementedError
1962+
1963+
19031964@delegate_names (SparseArray , ['npoints' , 'density' , 'fill_value' ,
19041965 'sp_values' ],
19051966 typ = 'property' )
1906- class SparseAccessor (PandasDelegate ):
1967+ class SparseAccessor (BaseAccessor , PandasDelegate ):
19071968 """
19081969 Accessor for SparseSparse from other sparse matrix data types.
19091970 """
19101971
1911- def __init__ (self , data = None ):
1912- self ._validate (data )
1913- # Store the Series since we need that for to_coo
1914- self ._parent = data
1915-
1916- @staticmethod
1917- def _validate (data ):
1972+ def _validate (self , data ):
19181973 if not isinstance (data .dtype , SparseDtype ):
1919- msg = "Can only use the '.sparse' accessor with Sparse data."
1920- raise AttributeError (msg )
1974+ raise AttributeError (self ._validation_msg )
19211975
19221976 def _delegate_property_get (self , name , * args , ** kwargs ):
1923- return getattr (self ._parent .values , name )
1977+ return getattr (self ._parent .array , name )
19241978
19251979 def _delegate_method (self , name , * args , ** kwargs ):
19261980 if name == 'from_coo' :
@@ -2034,3 +2088,188 @@ def to_coo(self, row_levels=(0, ), column_levels=(1, ), sort_labels=False):
20342088 column_levels ,
20352089 sort_labels = sort_labels )
20362090 return A , rows , columns
2091+
2092+ def to_dense (self ):
2093+ """
2094+ Convert a Series from sparse values to dense.
2095+
2096+ .. versionadded:: 0.25.0
2097+
2098+ Returns
2099+ -------
2100+ Series:
2101+ A Series with the same values, stored as a dense array.
2102+
2103+ Examples
2104+ --------
2105+ >>> series = pd.Series(pd.SparseArray([0, 1, 0]))
2106+ >>> series
2107+ 0 0
2108+ 1 1
2109+ 2 0
2110+ dtype: Sparse[int64, 0]
2111+
2112+ >>> series.sparse.to_dense()
2113+ 0 0
2114+ 1 1
2115+ 2 0
2116+ dtype: int64
2117+ """
2118+ from pandas import Series
2119+ return Series (self ._parent .array .to_dense (),
2120+ index = self ._parent .index ,
2121+ name = self ._parent .name )
2122+
2123+
2124+ class SparseFrameAccessor (BaseAccessor , PandasDelegate ):
2125+ """
2126+ DataFrame accessor for sparse data.
2127+
2128+ .. versionadded :: 0.25.0
2129+ """
2130+
2131+ def _validate (self , data ):
2132+ dtypes = data .dtypes
2133+ if not all (isinstance (t , SparseDtype ) for t in dtypes ):
2134+ raise AttributeError (self ._validation_msg )
2135+
2136+ @classmethod
2137+ def from_spmatrix (cls , data , index = None , columns = None ):
2138+ """
2139+ Create a new DataFrame from a scipy sparse matrix.
2140+
2141+ .. versionadded:: 0.25.0
2142+
2143+ Parameters
2144+ ----------
2145+ data : scipy.sparse.spmatrix
2146+ Must be convertible to csc format.
2147+ index, columns : Index, optional
2148+ Row and column labels to use for the resulting DataFrame.
2149+ Defaults to a RangeIndex.
2150+
2151+ Returns
2152+ -------
2153+ DataFrame
2154+ Each column of the DataFrame is stored as a
2155+ :class:`SparseArray`.
2156+
2157+ Examples
2158+ --------
2159+ >>> import scipy.sparse
2160+ >>> mat = scipy.sparse.eye(3)
2161+ >>> pd.DataFrame.sparse.from_spmatrix(mat)
2162+ 0 1 2
2163+ 0 1.0 0.0 0.0
2164+ 1 0.0 1.0 0.0
2165+ 2 0.0 0.0 1.0
2166+ """
2167+ from pandas import DataFrame
2168+
2169+ data = data .tocsc ()
2170+ index , columns = cls ._prep_index (data , index , columns )
2171+ sparrays = [
2172+ SparseArray .from_spmatrix (data [:, i ])
2173+ for i in range (data .shape [1 ])
2174+ ]
2175+ data = dict (zip (columns , sparrays ))
2176+ return DataFrame (data , index = index )
2177+
2178+ def to_dense (self ):
2179+ """
2180+ Convert a DataFrame with sparse values to dense.
2181+
2182+ .. versionadded:: 0.25.0
2183+
2184+ Returns
2185+ -------
2186+ DataFrame
2187+ A DataFrame with the same values stored as dense arrays.
2188+
2189+ Examples
2190+ --------
2191+ >>> df = pd.DataFrame({"A": pd.SparseArray([0, 1, 0])})
2192+ >>> df.sparse.to_dense()
2193+ A
2194+ 0 0
2195+ 1 1
2196+ 2 0
2197+ """
2198+ from pandas import DataFrame
2199+
2200+ data = {k : v .array .to_dense ()
2201+ for k , v in compat .iteritems (self ._parent )}
2202+ return DataFrame (data ,
2203+ index = self ._parent .index ,
2204+ columns = self ._parent .columns )
2205+
2206+ def to_coo (self ):
2207+ """
2208+ Return the contents of the frame as a sparse SciPy COO matrix.
2209+
2210+ .. versionadded:: 0.20.0
2211+
2212+ Returns
2213+ -------
2214+ coo_matrix : scipy.sparse.spmatrix
2215+ If the caller is heterogeneous and contains booleans or objects,
2216+ the result will be of dtype=object. See Notes.
2217+
2218+ Notes
2219+ -----
2220+ The dtype will be the lowest-common-denominator type (implicit
2221+ upcasting); that is to say if the dtypes (even of numeric types)
2222+ are mixed, the one that accommodates all will be chosen.
2223+
2224+ e.g. If the dtypes are float16 and float32, dtype will be upcast to
2225+ float32. By numpy.find_common_type convention, mixing int64 and
2226+ and uint64 will result in a float64 dtype.
2227+ """
2228+ try :
2229+ from scipy .sparse import coo_matrix
2230+ except ImportError :
2231+ raise ImportError ('Scipy is not installed' )
2232+
2233+ dtype = find_common_type (self ._parent .dtypes )
2234+ if isinstance (dtype , SparseDtype ):
2235+ dtype = dtype .subtype
2236+
2237+ cols , rows , datas = [], [], []
2238+ for col , name in enumerate (self ._parent ):
2239+ s = self ._parent [name ]
2240+ row = s .array .sp_index .to_int_index ().indices
2241+ cols .append (np .repeat (col , len (row )))
2242+ rows .append (row )
2243+ datas .append (s .array .sp_values .astype (dtype , copy = False ))
2244+
2245+ cols = np .concatenate (cols )
2246+ rows = np .concatenate (rows )
2247+ datas = np .concatenate (datas )
2248+ return coo_matrix ((datas , (rows , cols )), shape = self ._parent .shape )
2249+
2250+ @property
2251+ def density (self ):
2252+ """
2253+ Ratio of non-sparse points to total (dense) data points
2254+ represented in the DataFrame.
2255+ """
2256+ return np .mean ([column .array .density
2257+ for _ , column in self ._parent .iteritems ()])
2258+
2259+ @staticmethod
2260+ def _prep_index (data , index , columns ):
2261+ import pandas .core .indexes .base as ibase
2262+
2263+ N , K = data .shape
2264+ if index is None :
2265+ index = ibase .default_index (N )
2266+ if columns is None :
2267+ columns = ibase .default_index (K )
2268+
2269+ if len (columns ) != K :
2270+ raise ValueError ('Column length mismatch: {columns} vs. {K}'
2271+ .format (columns = len (columns ), K = K ))
2272+ if len (index ) != N :
2273+ raise ValueError ('Index length mismatch: {index} vs. {N}'
2274+ .format (index = len (index ), N = N ))
2275+ return index , columns
0 commit comments