Фильтрация кадра данных в Python с конкретными требованиями

Question

Я бы хотел отфильтровать фрейм данных, и у меня возникли трудности с этим.

Мой фрейм данных выглядит так:

+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+------+------+-------------+------+
|  node  |   date   | isSetl | qual  | run | firm | acct | type | isCust | seg  | ec | cc | currency | lov | sov | isM  | pbc  |   spanReq   | anov |
+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+------+------+-------------+------+
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | USD      |     |     | MNT  | CORE |   124073.69 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | CNY      |     |     |      |      |       43480 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |      117750 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | USD      |     |     | INIT | CORE |   124073.69 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | CNY      |     |     |      |      |       43480 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |      117750 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | CNH      |     |     | MNT  | CORE |           0 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | CNY      |     |     |      |      |      986680 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |    28786701 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |       67790 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | CNH      |     |     | INIT | CORE |           0 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | CNY      |     |     |      |      |      986680 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |    28786701 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |       67790 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | HKD      |     |     | MNT  | CORE | 17381842.35 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |      245850 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |     2193000 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | HKD      |     |     | INIT | CORE | 17381842.35 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |      245850 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | USD      |     |     |      |      |     2193000 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   70 | S    | TRUE   | CUST |    |    | HKD      |     |     | MNT  | CORE |      163900 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   70 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |      163900 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   70 | S    | TRUE   | CUST |    |    | HKD      |     |     | INIT | CORE |      163900 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   70 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |      163900 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   80 | S    | TRUE   | CUST |    |    | HKD      |     |     | MNT  | CORE |    25733800 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   80 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |    25733800 |    0 |
| oReq   | 20190627 | TRUE   | final |   0 | FCG  |   80 | S    | TRUE   | CUST |    |    | HKD      |     |     | INIT | CORE |    25733800 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   80 | S    | TRUE   | CUST |    |    | HKD      |     |     |      |      |    25733800 |    0 |
+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+------+------+-------------+------+

Я бы хотел отфильтровать так, чтобы мне были нужны строки ниже INIT в столбце 'isM'

Вывод, который я хочу:

+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+-----+-----+----------+------+
|  node  |   date   | isSetl | qual  | run | firm | acct | type | isCust | seg  | ec | cc | currency | lov | sov | isM | pbc | spanReq  | anov |
+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+-----+-----+----------+------+
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | CNY      |     |     |     |     |    43480 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   10 | S    | TRUE   | CUST |    |    | USD      |     |     |     |     |   117750 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | CNY      |     |     |     |     |   986680 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | HKD      |     |     |     |     | 28786701 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   40 | S    | TRUE   | CUST |    |    | USD      |     |     |     |     |    67790 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | HKD      |     |     |     |     |   245850 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   60 | S    | TRUE   | CUST |    |    | USD      |     |     |     |     |  2193000 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   70 | S    | TRUE   | CUST |    |    | HKD      |     |     |     |     |   163900 |    0 |
| curReq | 20190627 | TRUE   | final |   0 | FCG  |   80 | S    | TRUE   | CUST |    |    | HKD      |     |     |     |     | 25733800 |    0 |
+--------+----------+--------+-------+-----+------+------+------+--------+------+----+----+----------+-----+-----+-----+-----+----------+------+

Как мне отфильтровать, чтобы я мог получить вывод как таковой?

Нужно некоторое руководство в этом.

Answer 1

Это должно сработать.Он добавляет два временных столбца к фрейму данных (df):

• temp, чтобы отслеживать строки, в которых столбец isM равен INIT.Эти строки будут позже удалены.

• temp_ism заполняет столбец isM, чтобы найти все строки, следующие за INIT.

  df = df.assign(temp=df['isM'].eq('INIT'), temp_ism=df['isM'].ffill())
  # Drop the first `INIT` rows (`~df['temp']`) but retain the following rows.
  result = df[df['temp_ism'].eq('INIT') & ~df['temp']].iloc[:, :-2]  # Drop the two temporary columns.
  df = df.iloc[:, :-2]  # Drop the two temporary columns from the original dataframe.