Pandas中的Where（）方法与其他Pd.Series(Where() method in Pandas with other a Pd.Series)

Pandas中的Where（）方法与其他Pd.Series(Where() method in Pandas with other a Pd.Series)

我正在尝试pandas中的where（）方法。 我在文档页面中运行了一个简单的例子，其他的是pd.Series，我得到了NaN，我无法解释：

示例数据框是：

df = pd.DataFrame(np.arange(10).reshape(-1, 2), columns=['A', 'B'])

where（）子句是：

m = df % 3 == 0 n = pd.Series([100, 200]) df.where(m, n, axis = 1)

该方法返回以下数据帧：

A B 0 0.0 NaN 1 NaN 3.0 2 NaN NaN 3 6.0 NaN 4 NaN 9.0

我期待在A中看到100，在B中看到200而不是NaN。

你能解释一下NaN吗？ 您的建议将不胜感激。

I am experimenting with the where() method in pandas. I run the simple example in the documentation page with other being a pd.Series and I got NaNs which I can not explain:

The example dataframe is:

df = pd.DataFrame(np.arange(10).reshape(-1, 2), columns=['A', 'B'])

The where() clause is:

m = df % 3 == 0 n = pd.Series([100, 200]) df.where(m, n, axis = 1)

The method returns the following dataframe:

A B 0 0.0 NaN 1 NaN 3.0 2 NaN NaN 3 6.0 NaN 4 NaN 9.0

I was expecting to see 100 in A and 200 in B instead of NaNs.

Could you explain the NaNs? Your advice will be appreciated.

最满意答案

您的系列n没有合适的标签：

n Out: 0 100 1 200 dtype: int64

如果您将此作为other参数使用，则仅当索引相同时（在名为0和1的列上），它才会使用此Series。 如果找不到它们，它将显示NaN的。 但是，如果您更改标签：

n.index = ['A', 'B'] n Out: A 100 B 200 dtype: int64

现在它将按预期工作：

df.where(m, n, axis = 1) Out: A B 0 0 200 1 100 3 2 100 200 3 6 200 4 100 9

Your Series, n, does not have the appropriate labels:

n Out: 0 100 1 200 dtype: int64

If you use this as the other parameter, it will use this Series only when the index is the same (on columns named 0 and 1). If it cannot find them, it will display NaN's. However, if you change the labels:

n.index = ['A', 'B'] n Out: A 100 B 200 dtype: int64

Now it will work as you expect:

df.where(m, n, axis = 1) Out: A B 0 0 200 1 100 3 2 100 200 3 6 200 4 100 9