Pandas 创建DataFrame提示:type object ‘object‘ has no attribute ‘dtype‘

pandas版本0.25.3

import pandas as pd


symbol_info_columns = ['1', '持仓方向', '持仓量', '持仓收益率', '持仓收益', '持仓均价', '当前价格', '最大杠杆']  # v3
symbol_config = {'BTC': 'BTC-USDT-210924', 'LTC': 'LTC-USDT-210924', 'EOS': 'EOS-USDT-210924', 'ETH': 'ETH-USDT-210924', 'XRP': 'XRP-USDT-210924', 'FIL': 'FIL-USDT-210924'}
symbol_info = pd.DataFrame()
# dates = pd.date_range('20190101', periods=6)
# num_df = pd.DataFrame(data=np.random.randn(6, 8), index=dates, columns=symbol_info_columns)
symbol_info = pd.DataFrame(index=symbol_config.keys(), columns=symbol_info_columns)

data为空,且dtype默认为空时
出现type object ‘object’ has no attribute 'dtype’告警

Pandas 创建DataFrame提示:type object ‘object‘ has no attribute ‘dtype‘_第1张图片
原因分析:
创建DataFrame时,data字段为空
会默认创建一个空字典作为data

    def __init__(self, data=None, index=None, columns=None, dtype=None, copy=False):
        if data is None:
            data = {}

然后初始化字典

elif isinstance(data, dict):
    mgr = init_dict(data, index, columns, dtype=dtype)

init_dict函数中:
columns非空,且dtype默认为None时,会赋值nan_dtype = object

if columns is not None:
	if missing.any() and not is_integer_dtype(dtype):
	    if dtype is None or np.issubdtype(dtype, np.flexible):
	        # GH#1783
	        nan_dtype = object

该object下无dtype方法
可能是object引用错误

解决方案:
pandas(版本0.25.3)init_dict函数位于
D:\Users\。。。\Anaconda3\envs\Python3.7\Lib\site-packages\pandas\core\internals\construction.py

参考Python3.9环境中pandas(版本1.2.5)
同名函数(D:\Users\。。。\Anaconda3\envs\Python3.7\Lib\site-packages\pandas\core\internals\construction.py)写法

nan_dtype = np.dtype(object)

可见该问题应该是pandas(版本0.25.3)的bug

你可能感兴趣的:(pandas)