字典特征抽取

字典特征抽取，数据集类别特征比较多，需要将数据转换为字典，利用DictVectorizer进行字典特征抽取

# 导入模块
from sklearn.feature_extraction import DictVectorizer

data=[
    {'city':'BeiJing','temperature':100},
    {'city':'ShangHai','temperature':60},
    {'city':'ShenZheng','temperature':30}
]

transer = DictVectorizer(sparse=True) # sparse=True,稀疏矩阵，将非零的值的位置表示出来，节省内存，提高运行效率
data_new = transer.fit_transform(data)
print(data_new)
print(type(data_new))

  (0, 0)	1.0
  (0, 3)	100.0
  (1, 1)	1.0
  (1, 3)	60.0
  (2, 2)	1.0
  (2, 3)	30.0
<class 'scipy.sparse.csr.csr_matrix'>

transer = DictVectorizer(sparse=False) # sparse=True,稀疏矩阵，将非零的值的位置表示出来，节省内存，提高运行效率
data_new = transer.fit_transform(data)
print(data_new)
print(type(data_new))

[[  1.   0.   0. 100.]
 [  0.   1.   0.  60.]
 [  0.   0.   1.  30.]]
<class 'numpy.ndarray'>

print(transer.get_feature_names())

['city=BeiJing', 'city=ShangHai', 'city=ShenZheng', 'temperature']