python - 在 Pandas 中将分类特征转换为数字特征时出错

Question

我的数据框包含一个分类特征“Street”，它可以采用 2 个可能值“Grvl”或“Pave”中的一个。我试图在拟合机器学习算法之前将这个分类特征转换为数值。我的代码如下所示

dataset['Street']=dataset['Street'].map({'Grvl':0,'Pave':1}).astype(int)

我已用数据框中最常出现的值填充缺失值

dataset['Street'].isnull().sum()

我收到以下错误

    ValueError                                Traceback (most recent call last)
<ipython-input-59-86f0b031335a> in <module>()
      2     print dataset['Street'].isnull().sum()
      3     #dataset['MSZoning'] = dataset['MSZoning'].map( {'A': 0, 'C': 1,'FV': 2,'I':3,'RH':4,'RL':5,'RP':6,'RM':7} ).astype(int)
----> 4     dataset['Street']=dataset['Street'].map({'Grvl':0,'Pave':1}).astype(int)
      5     dataset['LotShape']=dataset['LotShape'].map({'Reg':0,'IR1':1,'IR2':2,'IR3':3}).astype(int)
      6     dataset['LandContour']=dataset['LandContour'].map({'Lvl':0,'Bnk':1,'HLS':2,'Low':3}).astype(int)

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\generic.pyc in astype(self, dtype, copy, raise_on_error, **kwargs)
   2948 
   2949         mgr = self._data.astype(dtype=dtype, copy=copy,
-> 2950                                 raise_on_error=raise_on_error, **kwargs)
   2951         return self._constructor(mgr).__finalize__(self)
   2952 

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\internals.pyc in astype(self, dtype, **kwargs)
   2936 
   2937     def astype(self, dtype, **kwargs):
-> 2938         return self.apply('astype', dtype=dtype, **kwargs)
   2939 
   2940     def convert(self, **kwargs):

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\internals.pyc in apply(self, f, axes, filter, do_integrity_check, consolidate, raw, **kwargs)
   2888 
   2889             kwargs['mgr'] = self
-> 2890             applied = getattr(b, f)(**kwargs)
   2891             result_blocks = _extend_blocks(applied, result_blocks)
   2892 

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\internals.pyc in astype(self, dtype, copy, raise_on_error, values, **kwargs)
    432                **kwargs):
    433         return self._astype(dtype, copy=copy, raise_on_error=raise_on_error,
--> 434                             values=values, **kwargs)
    435 
    436     def _astype(self, dtype, copy=False, raise_on_error=True, values=None,

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\internals.pyc in _astype(self, dtype, copy, raise_on_error, values, klass, mgr, **kwargs)
    475 
    476                 # _astype_nansafe works fine with 1-d only
--> 477                 values = com._astype_nansafe(values.ravel(), dtype, copy=True)
    478                 values = values.reshape(self.shape)
    479 

C:\Users\JAYASHREE\Anaconda2\lib\site-packages\pandas\core\common.pyc in _astype_nansafe(arr, dtype, copy)
   1912 
   1913         if np.isnan(arr).any():
-> 1914             raise ValueError('Cannot convert NA to integer')
   1915     elif arr.dtype == np.object_ and np.issubdtype(dtype.type, np.integer):
   1916         # work around NumPy brokenness, #1987

ValueError: Cannot convert NA to integer

Answer 1

您的数据框中有 NaN 值!由于如果存在缺失值，您无法将系列从对象转换为整数(使用 asType(int))，因此您应该在之前填充缺失值! 数据集['Street'].isnull().sum() 不填充缺失值

你可以使用pandas.DataFrame.fillna来做到这一点或sklearn.preprocessing.Imputer