DataFrame 按索引条件拆分。 将 DataFrame 拆分成条形

Question

I need to split Python DataFrame with datetime index column into Bars with specific "Length". Ie one dataframe should be splitted into 30min bars, 60min bars, 240 min bars and etc. Currently I'm trying to split it with loop like this:

class Bar:
    def __init__(self, o, h, l, c, v):
        self.Open = o
        self.High = h
        self.Low = l
        self.Close = c
        self.Volume = v
def GenerateBar(smallBars):
    o = smallBars["Open"][0]
    h = smallBars["High"].max()
    l = smallBars["Low"].min()
    c = smallBars["Close"][-1]
    vol = smallBars["Volume"].sum()
    return Bar(o, h, l, c, vol)
def GenerateBars(df, duration):
    while cur_time <= last_date:
        next_time = cur_time + timedelta(minutes = duration)
        cur_bar = GenerateBar(df[(df.index >= cur_time)&(df.index < next_time)])

But this way takes a way too long to split DataFrame into bars.

Answer 1

If you have a dataframe with a datetimeIndex, you can use resample . Also check out this guide .

Assuming df is your dataframe you want to convert, a resample for 30 minutes (30T) would look like this:

df = df.resample('30T').sum()

With this simple solution you have a problem if you need the mean of one column (eg price) and the sum of another column (eg volume). You can do this by using df.agg .

df = df.resample('30T').agg(['mean', 'sum'])