如何为Pandas DataFrame的列添加元数据？（价格指标开发场景）

方法1：利用DataFrame的attrs属性存储列元数据 #

Pandas的DataFrame和Series自带attrs属性，可直接存储自定义元数据。你可以给整个DataFrame的attrs添加一个字典，用列名作为键，对应的元数据（数据源、计算逻辑等）作为值，实现统一管理。

示例实现 #

import pandas as pd

# 初始化元数据字典
column_metadata = {
    "datetime": {"type": "原始数据", "description": "时间戳"},
    "open": {"type": "原始数据", "description": "开盘价"},
    "close": {"type": "原始数据", "description": "收盘价"},
}

# 绑定到DataFrame
price_df = pd.DataFrame(...)  # 你的原始数据
price_df.attrs["column_metadata"] = column_metadata

# 在指标计算函数中更新元数据
def average_day_range(price_df: pd.DataFrame, n: int=14, calculation_tool: int=0):
    # 原有计算逻辑保持不变
    function_dict = {
        0: {"func": sma, "name": "SMA"},
        1: {"func": ema, "name": "EMA"},
        # 其他计算方法...
    }
    function, name = function_dict[calculation_tool]["func"], function_dict[calculation_tool]["name"]
    high_var = function(price_df=price_df, input_mode=3, n=n, from_price=True)
    low_var = function(price_df=price_df, input_mode=4, n=n, from_price=True)
    adr = high_var[f'{name}_{n}'] - low_var[f'{name}_{n}']
    
    # 定义新列名并更新元数据
    col_name = f'Average Day Range {name}{n}'
    adr.rename(columns={0: col_name}, inplace=True)
    
    if "column_metadata" not in price_df.attrs:
        price_df.attrs["column_metadata"] = {}
    price_df.attrs["column_metadata"][col_name] = {
        "type": "计算指标",
        "calculation_source": ["high", "low"],
        "calculation_method": f"{name}({n})高值减去{name}({n})低值",
        "function": "average_day_range",
        "params": {"n": n, "calculation_tool": calculation_tool}
    }
    
    return adr

# 后续查询某列元数据
print(price_df.attrs["column_metadata"]["Average Day Range SMA14"])

方法2：用MultiIndex列名嵌入关键元数据 #

如果不想额外维护字典，可以将列名设为多层索引（MultiIndex），把元数据的核心信息（如类型、数据源）放在不同层级，既简洁又能直接识别列的属性。

示例实现 #

# 转换原始列为多层索引
price_df.columns = pd.MultiIndex.from_tuples([
    ("原始数据", "datetime"),
    ("原始数据", "open"),
    ("原始数据", "close"),
    ("原始数据", "high"),
    ("原始数据", "low"),
    ("指标", "indicator1"),
    ("指标", "indicator2")
])

# 在指标函数中用多层命名新列
def average_day_range(price_df: pd.DataFrame, n: int=14, calculation_tool: int=0):
    # 原有计算逻辑不变
    function, name = function_dict[calculation_tool]["func"], function_dict[calculation_tool]["name"]
    high_var = function(price_df=price_df, input_mode=3, n=n, from_price=True)
    low_var = function(price_df=price_df, input_mode=4, n=n, from_price=True)
    adr = high_var[f'{name}_{n}'] - low_var[f'{name}_{n}']
    
    # 多层列名：(类型, 指标名, 数据源)
    col_name = ("指标", f"ADR_{name}{n}", "源:high/low")
    adr.rename(columns={0: col_name}, inplace=True)
    
    return adr

# 快速筛选列
price_df["指标"]  # 筛选所有指标列
price_df["指标"].filter(like="ADR")  # 筛选所有ADR相关列

方法3：关联JSON文件持久化元数据 #

如果元数据复杂且需要持久化存储，可以单独维护一个JSON文件，每次新增列时更新文件，使用时读取即可。

示例JSON结构（column_metadata.json） #

{
    "datetime": {
        "type": "原始数据",
        "description": "时间戳"
    },
    "Average Day Range SMA14": {
        "type": "计算指标",
        "calculation_source": ["high", "low"],
        "calculation_method": "SMA(14)高值减SMA(14)低值",
        "function": "average_day_range",
        "params": {"n":14, "calculation_tool":0}
    }
}

函数中更新JSON #

import json

def update_metadata(col_name, metadata, filepath="column_metadata.json"):
    try:
        with open(filepath, "r") as f:
            data = json.load(f)
    except FileNotFoundError:
        data = {}
    data[col_name] = metadata
    with open(filepath, "w") as f:
        json.dump(data, f, indent=4)

# 在average_day_range中调用
col_name = f'Average Day Range {name}{n}'
update_metadata(col_name, {
    "type": "计算指标",
    "calculation_source": ["high", "low"],
    "calculation_method": f"{name}({n})高值减去{name}({n})低值",
    "function": "average_day_range",
    "params": {"n": n, "calculation_tool": calculation_tool}
})

方法4：自定义DataFrame子类封装元数据管理 #

如果需要更系统化的元数据管理，可以创建自定义DataFrame子类，内置元数据的添加、查询方法，让逻辑更清晰。

示例实现 #

class IndicatorDataFrame(pd.DataFrame):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.column_metadata = {}
    
    def add_col_metadata(self, col_name, metadata):
        self.column_metadata[col_name] = metadata
    
    def get_col_metadata(self, col_name):
        return self.column_metadata.get(col_name, "无对应元数据")

# 使用自定义类
price_df = IndicatorDataFrame(your_raw_data)

# 在指标函数中添加元数据
col_name = f'Average Day Range {name}{n}'
price_df.add_col_metadata(col_name, {
    "type": "计算指标",
    "calculation_source": ["high", "low"],
    "calculation_method": f"{name}({n})高值减去{name}({n})低值"
})

# 查询元数据
price_df.get_col_metadata("Average Day Range SMA14")

内容的提问来源于stack exchange，提问作者Jakub Szurlej