Python Pandas 使用什么规则来生成视图和副本？

What rules does Pandas use to generate a view vs a copy?

提问by orome

I'm confused about the rules Pandas uses when deciding that a selection from a dataframe is a copy of the original dataframe, or a view on the original.

我对 Pandas 在决定从数据帧中选择是原始数据帧的副本还是原始数据帧的视图时使用的规则感到困惑。

If I have, for example,

如果我有，例如，

df = pd.DataFrame(np.random.randn(8,8), columns=list('ABCDEFGH'), index=range(1,9))

I understand that a queryreturns a copy so that something like

我知道 aquery返回一个副本，以便像

foo = df.query('2 < index <= 5')
foo.loc[:,'E'] = 40

will have no effect on the original dataframe, df. I also understand that scalar or named slices return a view, so that assignments to these, such as

对原始数据框没有影响，df. 我也明白标量或命名切片返回一个视图，以便对这些进行赋值，例如

df.iloc[3] = 70

or

或者

df.ix[1,'B':'E'] = 222

will change df. But I'm lost when it comes to more complicated cases. For example,

会改变df。但是当涉及到更复杂的情况时，我就迷失了。例如，

df[df.C <= df.B] = 7654321

changes df, but

变化df，但是

df[df.C <= df.B].ix[:,'B':'E']

does not.

才不是。

Is there a simple rule that Pandas is using that I'm just missing? What's going on in these specific cases; and in particular, how do I change all values (or a subset of values) in a dataframe that satisfy a particular query (as I'm attempting to do in the last example above)?

Pandas 使用的是否有一个简单的规则我只是想念？在这些特定情况下发生了什么；特别是，如何更改满足特定查询的数据帧中的所有值（或值的子集）（正如我在上面的最后一个示例中尝试做的那样）？

---

---

Note: This is not the same as this question; and I have read the documentation, but am not enlightened by it. I've also read through the "Related" questions on this topic, but I'm still missing the simple rule Pandas is using, and how I'd apply it to — for example —?modify the values (or a subset of values) in a dataframe that satisfy a particular query.

注意：这与这个问题不同；我已经阅读了文档，但并没有受到启发。我也通读了关于这个主题的“相关”问题，但我仍然缺少 Pandas 正在使用的简单规则，以及我如何将它应用于 - 例如 -？修改值（或值的子集） ) 在满足特定查询的数据框中。

采纳答案by Jeff

Here's the rules, subsequent override:

这是规则，后续覆盖：

• All operations generate a copy

• If inplace=Trueis provided, it will modify in-place; only some operations support this

• An indexer that sets, e.g. .loc/.iloc/.iat/.atwill set inplace.

• An indexer that gets on a single-dtyped object is almost always a view (depending on the memory layout it may not be that's why this is not reliable). This is mainly for efficiency. (the example from above is for .query; this will alwaysreturn a copy as its evaluated by numexpr)

• An indexer that gets on a multiple-dtyped object is always a copy.

• 所有操作生成一个副本

• 如果inplace=True提供，它将就地修改；只有一些操作支持这个

• 设置的索引器，例如.loc/.iloc/.iat/.at将设置到位。

• 获取单数据类型对象的索引器几乎总是一个视图（取决于内存布局，这可能不是这不可靠的原因）。这主要是为了效率。（上面的示例是 for .query；这将始终返回一个副本作为其评估的numexpr）

• 获取多类型对象的索引器始终是副本。

Your example of chained indexing

你的例子 chained indexing

df[df.C <= df.B].loc[:,'B':'E']

is not guaranteed to work (and thus you shoulld neverdo this).

不能保证工作（因此你永远不应该这样做）。

Instead do:

而是这样做：

df.loc[df.C <= df.B, 'B':'E']

as this is fasterand will always work

因为这更快并且将始终有效

The chained indexing is 2 separate python operations and thus cannot be reliably intercepted by pandas (you will oftentimes get a SettingWithCopyWarning, but that is not 100% detectable either). The dev docs, which you pointed, offer a much more full explanation.

链式索引是 2 个独立的 python 操作，因此不能被 Pandas 可靠地拦截（你经常会得到SettingWithCopyWarning，但这也不是 100% 可检测的）。您指出的dev docs提供了更完整的解释。