从Xarray中的“xr.Dataset”获取值(点、矢量、数组等)?(Python 3)

Question

我不知道如何从xr.Dataset对象中实际提取数据。我不知道如何获取个人价值观。如何像使用DataArrays一样从数据集中提取值（点值、向量、数组等）？

np.random.seed(0)
DA_data = xr.DataArray(np.random.random((3,2,10,100)), dims=["targets","accuracy","metrics","attributes"], name="Data")
DA_data.coords["attributes"] = ["attr_%d"%_ for _ in range(100)]
# DA_data.coords
# Coordinates:
#   * targets     (targets) int64 0 1 2
#   * accuracy    (accuracy) int64 0 1
#   * metrics     (metrics) int64 0 1 2 3 4 5 6 7 8 9
#   * attributes  (attributes) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 ...
# Indexing DataArray
#DA_data.sel(targets=0, accuracy=0, metrics=0, attributes="attr_5").values
#array(0.6458941130666561)
float(DA_data.sel(targets=0, accuracy=0, metrics=0, attributes="attr_5").values)
#0.6458941130666561
# Indexing Dataset
DS_data = DA_data.to_dataset()
# DS_data
# <xarray.Dataset>
# Dimensions:     (accuracy: 2, attributes: 100, metrics: 10, targets: 3)
# Coordinates:
#   * targets     (targets) int64 0 1 2
#   * accuracy    (accuracy) int64 0 1
#   * metrics     (metrics) int64 0 1 2 3 4 5 6 7 8 9
#   * attributes  (attributes) <U7 'attr_0' 'attr_1' 'attr_2' 'attr_3' ...
# Data variables:
#     Data        (targets, accuracy, metrics, attributes) float64 0.5488 ...
DS_data.sel(targets=0, accuracy=0, metrics=0, attributes="attr_5").values
# <bound method Mapping.values of <xarray.Dataset>
# Dimensions:     ()
# Coordinates:
#     targets     int64 0
#     accuracy    int64 0
#     metrics     int64 0
#     attributes  <U7 'attr_5'
# Data variables:
#     Data        float64 0.6459>
float(DS_data.sel(targets=0, accuracy=0, metrics=0, attributes="attr_5").values)
# ---------------------------------------------------------------------------
# TypeError                                 Traceback (most recent call last)
# <ipython-input-408-e0c88e8541d8> in <module>()
#      38 # Data variables:
#      39 #     Data        float64 0.6459>
# ---> 40 float(DS_data.sel(targets=0, accuracy=0, metrics=0, attributes="attr_5").values)
# TypeError: float() argument must be a string or a number, not 'method'

Answer 1

这有点令人困惑，但.values在Dataset和DataArray:上的工作方式不同

• DataArray.values返回一个NumPy数组。这种行为与熊猫是一致的
• Dataset.values()返回组成数据集的DataArray对象的列表（Python 2）或ValuesView（Python 3）。这种行为与满足Python的Mapping接口的Dataset一致

要从Dataset中提取值，需要通过数据集类似字典的接口（例如float(DA_data['Data'])或float(DA_data.values()[0])）提取DataArray。您不能直接将Dataset转换为float或NumPy数组，就像使用Python dict一样。