python-3.x - 重采样数据帧的 2D 插值

Question

我想在重新采样后对数据帧进行 2D 插值。重采样部分可以是按天、按月或按分钟。下面的例子只是为了说明这个过程。我不想从数据帧中逐列删除数据，而是想在数据帧本身中执行重新采样和插值。

例子：

import pandas as pd
import scipy.interpolate

df = pd.DataFrame.from_dict({'time': {0: pd.Timestamp('2015-01-01 00:30:00'),
  1: pd.Timestamp('2015-01-01 00:30:00'),
  2: pd.Timestamp('2015-01-01 00:30:00'),
  3: pd.Timestamp('2015-01-01 00:30:00'),
  4: pd.Timestamp('2015-02-01 00:30:00'),
  5: pd.Timestamp('2015-02-01 00:30:00'),
  6: pd.Timestamp('2015-02-01 00:30:00'),
  7: pd.Timestamp('2015-02-01 00:30:00')},
 'x': {0: 25.0,
  1: 25.0,
  2: 25.5,
  3: 25.5,
  4: 25.0,
  5: 25.0,
  6: 25.5,
  7: 25.5},
 'y': {0: 51.25,
  1: 51.875,
  2: 51.25,
  3: 51.875,
  4: 51.25,
  5: 51.875,
  6: 51.25,
  7: 51.875},
 'z': {0: 1.9466316037407283e-13,
  1: 1.4071156620529962e-13,
  2: 2.1328328914192035e-13,
  3: 1.5983803094022503e-13,
  4: 1.8673746631035076e-13,
  5: 1.1844342238769012e-13,
  6: 2.0592551372867335e-13,
  7: 1.341905238186683e-13}})
df.set_index("time", inplace=True)

然后插值片：

interp_x = df["x"].values
interp_y = df["y"].values
interp_z = df["z"].values
interp_f = scipy.interpolate.interp2d(x=interp_x, y=interp_y, z=interp_z)

最后，以下

df.resample("1M").apply(lambda r: interp_f(r["x"], r["y"]))

没有按预期工作，因为我得到：

time
2015-01-31    [[1.9070031334221175e-13, 1.9070031334221175e-...
2015-02-28    [[1.9070031334221175e-13, 1.9070031334221175e-...
Freq: M, dtype: object

标签： python-3.xpandasinterpolation