“Pandas”列没有值

Question

因此，我试图为一个数据帧创建一个新的列，当mfi超过70时，该列本质上有一个1，当mfi超过70时，它有一个0。到目前为止，守则是：

import pandas as pd
import numpy as np

#get stock prices
d = pd.read_csv(r"C:\Users\B1880\Downloads\AMD_stock_data\AMD_2020_2020.txt")
d.columns = ['Dates', 'Open', 'High', 'Low', 'Close', 'Volume']
d.set_index(d['Dates'], inplace=True)
d.drop(['Dates'], axis=1, inplace=True)

#MONEY FLOW INDEX
d['typical_price'] = (d['High'] + d['Low'] + d['Close'])/3 
d['raw_money_flow'] = d['typical_price']*d['Volume']
mf = d.raw_money_flow.diff(1) 
p = mf.copy()
n = mf.copy()
p[p<=0] = 0
n[n>0] = 0
pmf = p.rolling(window=14).mean()
nmf = abs(n.rolling(window=14).mean())
mfr = pmf / nmf
d['mfi'] = 100 - (100 / (mfr +1))
d['mfi'].dropna(inplace=True)

# # #mfi location
d['mfi_70_overbought'] = np.where(d['mfi'] > 70, 1, 0)
d['mfi_70_overbought']

当我像这样运行代码时，我得到了错误ValueError: Length of values does not match length of index，为了解决这个问题，我做了d['mfi_70_overbought'] = pd.Series(np.where(d['mfi'] > 70, 1, 0))。虽然现在当我打印d['mfi_70_overbought']列时，整个列都被NAN值填充。鉴于mfi的值肯定超过70，我遗漏了什么？谢谢大家!

编辑：这是d['mfi']打印输出的内容：

Dates
2010-01-04 07:18:00          NaN
2010-01-04 07:23:00          NaN
2010-01-04 07:29:00          NaN
2010-01-04 07:38:00          NaN
2010-01-04 07:44:00          NaN
                         ...    
2019-12-31 19:55:00    54.775561
2019-12-31 19:56:00    49.240351
2019-12-31 19:57:00    54.346136
2019-12-31 19:58:00    86.883785
2019-12-31 19:59:00    50.210623
Name: mfi, Length: 1293557, dtype: float64

数据的URL为：https://docs.google.com/spreadsheets/d/1uxVjEJkEmDZwu44pNxsg5ZBonqbTFak8HoESbxo0AM0/edit?usp=sharing

Answer 1

# necessary imports
import pandas as pd
import numpy as np

---

设置

试图重现你所做的

模拟数据：

data = {'timestep1': [45,46,47,48,1000],
        'timestep2': [46,47,48,49,2020],
        'timestep3': [47,48,49,50,1002],
        'timestep4': [50,49,48,47, 99],
        'timestep5': [45,40,50,70,2500]}

名称列，设置索引：

df = pd.DataFrame.from_dict(data, orient='index')
df.columns = ['Open', 'High', 'Low', 'Close', 'Volume']
df.index.name = 'Dates'

计算：

df['typical_price'] = (df['High'] + df['Low'] + df['Close'])/3 
df['raw_money_flow'] = df['typical_price']*df['Volume']
mf = df.raw_money_flow.diff(1)

p = mf.copy()
n = mf.copy()    
p[p<=0] = 0
n[n>0] = 0

windowsize=2  # example value
pmf = p.rolling(window=windowsize).mean()
nmf = abs(n.rolling(window=windowsize).mean())     
mfr = pmf/nmf

df['mfi'] = 100 - (100 / (mfr +1))    
df['mfi'].dropna(inplace=True)

---

问题

现在如果我运行df['mfi_70_overbought'] = np.where(df['mfi'] > 70, 1, 0)，我会得到相同的错误：ValueError: Length of values does not match length of index

---

解决方案

如果您只想在mfi超过70时拥有一个1的新列，而不是时拥有一个0的新列，那么您可以避免numpy并使用pandas工具

定义一个函数，如果其输入大于70，则返回1，否则应返回0：

def above70(num):
    return int(num > 70)

Apply这对df[mfi]来说：

df['mfi'].apply(above70)

在我的示例中，此新列将如下所示：

Dates
timestep3    0
timestep4    0
timestep5    1
Name: mfi, dtype: int64

---

附带问题

这个新列比原始数据帧的列短（差异为windowsize），因为之前我们应用了^{}和^{}。如果要将此列附加到数据帧，请填充此列，或者不执行缩短此列的步骤