如何在Pandas中按特定条件将Result行切片为多个子DataFrame？

问题描述 #

需要将Pandas DataFrame中Step列为"Result"的行，以Point列值为"P1"且X_Y列值为"X"的位置作为起始点，分割为多个子DataFrame并存入list_of_dfs列表。尝试定位所有"P1"的索引后用列表推导式切片，但得到的列表包含两个空DataFrame，寻求解决方法。

用户原代码 #

import pandas as pd

df = pd.DataFrame(
    {
        "Step": (
            "1", "1", "1", "1", "1", "2", "2", "2", "2", "2", "Result", "Result", "Result", "Result", "Result",
            "1", "1", "1", "1", "1", "2", "2", "2", "2", "2", "Result", "Result", "Result", "Result", "Result"
        ),
        "Point": (
            "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3",
            "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3",
        ),
        "X_Y": (
            "X", "X", "Y", "X", "Y",  "X", "X", "Y", "X", "Y", "X", "X", "Y", "X", "Y", 
            "X", "X", "Y", "X", "Y",  "X", "X", "Y", "X", "Y", "X", "X", "Y", "X", "Y",
        ),
        "Value A": (
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 
        ),
        "Value B": (
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
        ),
    }
)

dff = df.loc[df["Step"] == "Result"]

value = "P1"
tuple_of_positions = list()

result = dff.isin([value])

seriesObj = result.any()
columnNames = list(seriesObj[seriesObj == True].index)

for col in columnNames:
    rows = list(result[col][result[col] == True].index)
    for row in rows:
        tuple_of_positions.append((row, col))

length_of_one_df = (len(dff["Point"].unique().tolist()) * 2 ) - 1

list_of_dfs = [dff.iloc[x : x + length_of_one_df] for x in rows]

print(list_of_dfs)

问题分析 #

原代码存在两个核心问题：

1. 最终切片用的rows变量是循环最后一次赋值的结果，并非所有符合Point="P1"且X_Y="X"的索引位置；
2. 使用iloc时传入了原DataFrame的索引标签，但iloc需要的是相对行号，原索引值超出dff的行范围后，就会产生空DataFrame。

解决方法 #

修正后的代码 #

import pandas as pd

df = pd.DataFrame(
    {
        "Step": (
            "1", "1", "1", "1", "1", "2", "2", "2", "2", "2", "Result", "Result", "Result", "Result", "Result",
            "1", "1", "1", "1", "1", "2", "2", "2", "2", "2", "Result", "Result", "Result", "Result", "Result"
        ),
        "Point": (
            "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3",
            "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3", "P1", "P2", "P2", "P3", "P3",
        ),
        "X_Y": (
            "X", "X", "Y", "X", "Y",  "X", "X", "Y", "X", "Y", "X", "X", "Y", "X", "Y", 
            "X", "X", "Y", "X", "Y",  "X", "X", "Y", "X", "Y", "X", "X", "Y", "X", "Y",
        ),
        "Value A": (
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 
        ),
        "Value B": (
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
            70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72, 70, 68, 66.75, 68.08, 66.72,
        ),
    }
)

# 筛选Step为Result的行
dff = df.loc[df["Step"] == "Result"]

# 精准筛选符合条件的起始行，并获取其在dff中的相对行号
start_mask = (dff["Point"] == "P1") & (dff["X_Y"] == "X")
start_positions = list(dff[start_mask].index.get_indexer(dff.index))

# 每个子DataFrame的长度，根据数据结构固定为5
length_of_one_df = 5

# 生成子DataFrame列表
list_of_dfs = [dff.iloc[x:x+length_of_one_df] for x in start_positions]

# 验证结果
for idx, sub_df in enumerate(list_of_dfs):
    print(f"子DataFrame {idx+1}:")
    print(sub_df)
    print("-"*30)

关键修正点 #

• 精准筛选起始行：用(dff["Point"] == "P1") & (dff["X_Y"] == "X")同时满足两个条件，避免误选其他含P1的行；
• 获取相对行号：通过index.get_indexer(dff.index)将原索引标签转换为dff内部的相对行号，确保iloc切片时使用正确的位置；
• 正确切片：从每个起始行号开始，取连续5行（对应一组完整的P1-P3的X/Y数据），得到两个非空的子DataFrame。

内容的提问来源于stack exchange，提问作者Bakira