熊猫根据条件组合连续的行

Question

我的问题与这个问题类似，但答案似乎并不完全有效！！

根据条件合并行熊猫数据框

鉴于以下熊猫数据框：

+---------+-----------------+-----------------+
| SECTION | TEXT            | NUMBER_OF_WORDS |
+---------+-----------------+-----------------+
| ONE     | lots   of text… | 55              |
+---------+-----------------+-----------------+
| ONE     | word1           | 1               |
+---------+-----------------+-----------------+
| ONE     | lots   of text… | 151             |
+---------+-----------------+-----------------+
| ONE     | word2           | 1               |
+---------+-----------------+-----------------+
| ONE     | word3           | 1               |
+---------+-----------------+-----------------+
| ONE     | word4           | 1               |
+---------+-----------------+-----------------+
| TWO     | lots   of text… | 523             |
+---------+-----------------+-----------------+
| TWO     | lots   of text… | 123             |
+---------+-----------------+-----------------+
| TWO     | word4           | 1               |
+---------+-----------------+-----------------+

如果 NUMBER_OF_WORDS 列中的值为 1； 它必须与上一行结合； 只要它们具有相同的 SECTION 值。

因此最终的结果应该是这样的：

+---------+--------------------------------------+-----------------+
| SECTION | TEXT                                 | NUMBER_OF_WORDS |
+---------+--------------------------------------+-----------------+
| ONE     | lots   of text…, word1               | 56              |
+---------+--------------------------------------+-----------------+
| ONE     | lots   of text…, word2, word3, word4 | 154             |
+---------+--------------------------------------+-----------------+
| TWO     | lots   of text…                      | 523             |
+---------+--------------------------------------+-----------------+
| TWO     | lots   of text…, word4               | 124             |
+---------+--------------------------------------+-----------------+

这是代码； 这似乎有效，但不是我想要的。

df.groupby(['SECTION', (df.NUMBER_OF_WORDS.shift(1) == 1)], as_index=False, sort=False).agg({'TEXT': lambda x: ', '.join(x), 'NUMBER_OF_WORDS': lambda x: sum(x)})

更新

这是来自 BEN_YO 的回答； 但他似乎有一个小错字。 为了让未来的用户回答这个问题，我将把他的答案稍微修改一下。

s = df['NUMBER_OF_WORDS'].ne(1).cumsum()
out = df.groupby(s).agg({'SECTION': 'first','TEXT': lambda x: ', '.join(x),'NUMBER_OF_WORDS': lambda x: sum(x)})

Answer 1

让我们用cumsum试试groupby

s = df['NUMBER_OF_WORDS'].ne(1).cumsum()
out = df.groupby(s).agg({'SECTION':'first','TEXT':','.join,'NUMBER_OF_WORDS':'sum'})