从嵌套字典构建数据框架的最佳方法

Question

处理完数据后，我在以下结构（嵌套字典）中保存了组级别计算：

{'Source1': {(1, 2): {'value1': -1.4089917877152731, 'value2': 0.15890127107708821}, (1, 3): {'value1': -3.6436438771179183, 'value2': 0.00027189114106343325}, (1, 4): {'value1': 1.3921379718956783, 'value2': 0.1639443047264573}, (2, 3): {'value1': -2.1272739953077449, 'value2': 0.033444556519261023}, (3, 4): {'value1': 5.0887284442498775, 'value2': 3.7318559307126006e-07}, (2, 4): {'value1': 2.781268059718232, 'value2': 0.0054326884405563099}}, 'Source2': {(1, 2): {'value1': 1.6065065530210021, 'value2': 0.10840303417258132}, (1, 3): {'value1': -0.67561007794063666, 'value2': 0.49941051115943469}, (1, 4): {'value1': -0.99500921260852215, 'value2': 0.31991568858488023}, (2, 3): {'value1': -2.4076869756909676, 'value2': 0.016168545874782416}, (3, 4): {'value1': -0.31851460166510093, 'value2': 0.75013768971795858}, (2, 4): {'value1': -2.7976881039916965, 'value2': 0.0052043800033575345}}}

字典的结构是：

Level1 - Source
Level2 - Group comparison (1,2) means compare Group 1 vs Group2
Level3 - Value of the comparison in two measurement types

我的目标是获得下表

Source Name | (1,2) | (1,3) | (1,4) | ......| (2,4)|
Source1 | value1 | value1 | ................| value1

• 我认为包含价值2在某种程度上是不可能的，所以忽略了价值2

什么是将其转换为熊猫数据框架的最佳方式。

Answer 1

我想你需要：

df = pd.concat({k:pd.DataFrame(v) for k, v in d.items()})
df.columns = ['({},{})'.format(i,j) for i,j in df.columns]
print (df)
                   (1,2)     (1,3)     (1,4)     (2,3)     (2,4)         (3,4)
Source1 value1 -1.408992 -3.643644  1.392138 -2.127274  2.781268  5.088728e+00
        value2  0.158901  0.000272  0.163944  0.033445  0.005433  3.731856e-07
Source2 value1  1.606507 -0.675610 -0.995009 -2.407687 -2.797688 -3.185146e-01
        value2  0.108403  0.499411  0.319916  0.016169  0.005204  7.501377e-01