python - 如何在 Pandas Dataframe 中增量添加行？

Question

我正在计算从 9:15 到 15:30 每 15 分钟持续时间内的数据开高低关 (OHLC)，并希望将 OHLC 值存储在每个新行的数据框中。

ohlc = pd.DataFrame(columns=('Open','High','Low','Close'))
for row in ohlc:
    ohlc.loc[10] = pd.DataFrame([[candle_open_price,candle_high_price,candle_low_price,candle_close_price]])

但我不能说得到以下错误:

ValueError: cannot set a row with mismatched columns

只是我想增量存储我计算的每 15 分钟持续时间的 OHLC 数据，并将其放入新的 ohlc 数据帧的行中

---

编辑

import numpy as np
import pandas as pd
import datetime as dt
import matplotlib as plt
import dateutil.parser

tradedata = pd.read_csv('ICICIBANK_TradeData.csv', index_col=False, 
              names=['Datetime','Price'], 
            header=0)
tradedata['Datetime'] =  pd.to_datetime(tradedata['Datetime'])

first_trd_time = tradedata['Datetime'][0]
last_time = dateutil.parser.parse('2016-01-01 15:30:00.000000')

candle_time = 15;
candle_number = 0

while(first_trd_time < last_time):
    candledata = tradedata[(tradedata['Datetime']>first_trd_time) & (tradedata['Datetime']<first_trd_time+dt.timedelta(minutes=candle_time))]
first_trd_time = first_trd_time+dt.timedelta(minutes=candle_time)

candle_open_price = candledata.iloc[0]['Price']
candle_open_time = candledata.iloc[0]['Datetime']
candle_close_price = candledata.iloc[-1]['Price']
candle_close_time = candledata.iloc[-1]['Datetime']
candle_high_price = candledata.loc[candledata['Price'].idxmax()]['Price']
candle_high_time = candledata.loc[candledata['Price'].idxmax()]['Datetime'] 
candle_low_price = candledata.loc[candledata['Price'].idxmin()]['Price']
candle_low_time = candledata.loc[candledata['Price'].idxmin()]['Datetime']

ohlc = pd.DataFrame(columns=('Open','High','Low','Close'))
ohlc_data = pd.DataFrame()

if(candle_number == 0):
    ohlc = pd.DataFrame(np.array([[0, 0, 0, 0]]), columns=['Open', 'High', 'Low', 'Close']).append(ohlc, ignore_index=True)
    candle_number = candle_number + 1
    print "Zeroth Candle"
else:
    ohlc.ix[candle_number] = (candle_open_price,candle_open_price,candle_open_price,candle_open_price)
    print "else part with incermenting candle_number"
    candle_number = candle_number + 1

print "first_trd_time" 
print first_trd_time
print candle_number

print "Success!"

这是我的代码错误是

ValueError: cannot set by positional indexing with enlargement

Answer 1

IIUC 您可以将每一行的 DataFrames append 到 DataFrames dfs 列表，然后是 concat他们到 df1:

ohlc = pd.DataFrame(columns=('Open','High','Low','Close'))

dfs = []
for row in ohlc.iterrows():
    df = pd.DataFrame([candle_open_price,candle_high_price,
                        candle_low_price,candle_close_price]).T
    dfs.append(df)

df1 = pd.concat(dfs, ignore_index=True)
print (df1)

然后concat到原来的DataFrame ohlc:

df2 = pd.concat([ohlc,df1])
print (df2)

---

示例(用于在循环的每次迭代中添加相同数据的测试):

#sample data
candle_open_price = pd.Series([1.5,10], 
                              name='Open', 
                              index=pd.DatetimeIndex(['2016-01-02','2016-01-03']) )
candle_high_price =  pd.Series([8,9], 
                               name='High', 
                               index=pd.DatetimeIndex(['2016-01-02','2016-01-03']))
candle_low_price =  pd.Series([0,12], 
                              name='Low', 
                              index=pd.DatetimeIndex(['2016-01-02','2016-01-03']))
candle_close_price =  pd.Series([4,5], 
                                name='Close', 
                                index=pd.DatetimeIndex(['2016-01-02','2016-01-03']))

data = np.array([[1,2,3,5],[7,7,8,9],[10,8,9,3]])
idx = pd.DatetimeIndex(['2016-01-08','2016-01-09','2016-01-10'])
ohlc = pd.DataFrame(data=data, 
                    columns=('Open','High','Low','Close'),
                    index=idx)
print (ohlc)
            Open  High  Low  Close
2016-01-08     1     2    3      5
2016-01-09     7     7    8      9
2016-01-10    10     8    9      3

dfs = []
for row in ohlc.iterrows():
    df = pd.DataFrame([candle_open_price,candle_high_price,
                       candle_low_price,candle_close_price]).T
    #print (df)
    dfs.append(df)

df1 = pd.concat(dfs)
print (df1)
            Open  High   Low  Close
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0

df2 = pd.concat([ohlc,df1])
print (df2)
            Open  High   Low  Close
2016-01-08   1.0   2.0   3.0    5.0
2016-01-09   7.0   7.0   8.0    9.0
2016-01-10  10.0   8.0   9.0    3.0
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0
2016-01-02   1.5   8.0   0.0    4.0
2016-01-03  10.0   9.0  12.0    5.0