根据 DF 中更改的不同字符串列停止前向填充

Question

我是 Python 的新手，所以放轻松！

我有一个如下所示的数据框。我想向前填充 shares_owned 列中的 NaN，但当 df['ticker'] 中的字符串更改时停止。并且仅当另一个数字再次出现在 shares_owned 中时才开始。

date	ticker	shares_owned	price
01/01/2020	EZY	NaN	£2
02/01/2020	EZY	10	£2.1
03/01/2020	EZY	NaN	£2.12
04/01/2020	EZY	NaN	£12.5
01/01/2020	FTSE	NaN	£11
02/01/2020	FTSE	NaN	£12
03/01/2020	FTSE	2	£12.5
04/01/2020	FTSE	NaN	£12.5

例如，输出 table 将如下所示：

date	ticker	shares_owned	price
01/01/2020	EZY	NaN	£2
02/01/2020	EZY	10	£2.1
03/01/2020	EZY	10	£2.12
04/01/2020	EZY	10	£12.5
01/01/2020	FTSE	NaN	£11
02/01/2020	FTSE	NaN	£12
03/01/2020	FTSE	2	£12.5
04/01/2020	FTSE	2	£12.5

到目前为止，我一直在尝试使用 .fillna(method='ffill') 无济于事。

Answer 1

你注意到分组，因此 groupby() 进行分组
组内fillna(method="fill")组内transform()

df = pd.read_csv(io.StringIO("""date    ticker  shares_owned    price
01/01/2020  EZY NaN £2
02/01/2020  EZY 10  £2.1
03/01/2020  EZY NaN £2.12
04/01/2020  EZY NaN £12.5
01/01/2020  FTSE    NaN £11
02/01/2020  FTSE    NaN £12
03/01/2020  FTSE    2   £12.5
04/01/2020  FTSE    NaN £12.5"""), sep="\t")

df["shares_owned"] = df.groupby("ticker")["shares_owned"].transform(lambda s: s.fillna(method="ffill"))

输出

	date	ticker	shares_owned	price
0	01/01/2020	EZY	nan	£2
1	02/01/2020	EZY	10	£2.1
2	03/01/2020	EZY	10	£2.12
3	04/01/2020	EZY	10	£12.5
4	01/01/2020	FTSE	nan	£11
5	02/01/2020	FTSE	nan	£12
6	03/01/2020	FTSE	2	£12.5
7	04/01/2020	FTSE	2	£12.5

根据 DF 中更改的不同字符串列停止前向填充

Stopping a forward fill based on a different string column changing in the DF

python

pandas

dataframe

fillna

输出