Pandas Dataframe按照给定的index列表切分成多个小Dataframe

整个思路比较简单：

1. 根据dataframe.loc[]来框一个子dataframe
2. 由于.loc[]是左闭右闭，所以如果希望哪一边开的话，需要删除哪一边的数据

使用的工具函数如下：

def split_dataframe_by_index_list(dataframe: pd.DataFrame, index_list: list, left=False, right=False):
    """按照index的列表切分dataframe
    :param dataframe: 待切分的dataframe
    :param index_list: 切分的index的列表，注意需要相对于dataframe的index是有序的
    :param left: 是否包含左边的值
    :param right: 是否包含右边的值
    """
    if len(index_list) == 0:
        yield dataframe
    else:
        if dataframe.index.values[0] not in index_list:
            index_list.insert(0, dataframe.index.values[0])
        if dataframe.index.values[-1] not in index_list:
            index_list.append(dataframe.index.values[-1])
        for index in range(len(index_list) - 1):
            this_df: pd.DataFrame = dataframe.loc[index_list[index]:index_list[index + 1], :]
            if not left:
                this_df = this_df.drop(index_list[index], axis=0)
            if not right:
                this_df = this_df.drop(index_list[index + 1], axis=0)
            if this_df.shape[0] > 1:
                yield this_df

示例代码

import pandas as pd
import numpy as np


def split_dataframe_by_index_list(dataframe: pd.DataFrame, index_list: list, left=False, right=False):
    """按照index的列表切分dataframe
    :param dataframe: 待切分的dataframe
    :param index_list: 切分的index的列表，注意需要相对于dataframe的index是有序的
    :param left: 是否包含左边的值
    :param right: 是否包含右边的值
    """
    if len(index_list) == 0:
        yield dataframe
    else:
        if dataframe.index.values[0] not in index_list:
            index_list.insert(0, dataframe.index.values[0])
        if dataframe.index.values[-1] not in index_list:
            index_list.append(dataframe.index.values[-1])
        for index in range(len(index_list) - 1):
            this_df: pd.DataFrame = dataframe.loc[index_list[index]:index_list[index + 1], :]
            if not left:
                this_df = this_df.drop(index_list[index], axis=0)
            if not right:
                this_df = this_df.drop(index_list[index + 1], axis=0)
            if this_df.shape[0] > 1:
                yield this_df


if __name__ == '__main__':
    df = pd.DataFrame(data=np.random.randint(0, 100, size=(50, 3)))
    for sub_df in split_dataframe_by_index_list(df, [11, 22, 33, 44]):
        print(sub_df.shape)

效果：

(10, 3) 1 10
(10, 3) 12 21
(10, 3) 23 32
(10, 3) 34 43
(4, 3) 45 48