CODEWITHSUNDEEP
pythonpandasresampling
Jitesh
Jitesh

Reputation: 11

Resample pandas time series 30 mins for 9:15 as start time

I am trying to resample OHLC data to 30 mins. The market data starts at at 9:15 and I would like the resampled time to have 9:15-9:45 and so on. But I am able to get the data resampled as 9:00-9:30

Paste Bin link to 1 min market data

pd.DataFrame(download_data).set_index('date'['close'].resample('30T').ohlc()

As you see in the picture the start time is 9:00 and not 9:15...

Upvotes: 1

Views: 3696

Answers (5)

marizombie
marizombie

Reputation: 453

I will summarize the answers and will add another option for origin. So yeah, resample method had and argument base as a starting position, so it was possible to write it like:

data.resample('24h', base=9.5).ohlc()

I am using random data just to show how it is done. The base takes floats, so I assumed 9.5 would be 9:30 and it seems to work. So I am saying it was possible to use the base argument because it is deprecated already. Although it still brings nice output: enter image description here

As for now, there is a new, nice and clear argument named origin in resample method. It takes string as an input, as other answer mentions it also takes keywords like 'start' or 'end' but also a string date, so we just put here something something 9:30 and here is what we get:

data.resample('24h', origin='2011-12-31 09:30:00').ohlc()

And here is the output:

enter image description here

So just try it with your data.

Upvotes: 0

Avnish Midha
Avnish Midha

Reputation: 11

@TFA the use of origin='start' may not be that useful esp. if our start time does not fit in that rule of starting from 9:15 am, e.g. if we are trying to convert 1 min data to 15 mins, and we give data from 9:37 am, the 15mins candles will start from 9:37 am, which is not what we need.

The better solution that I could find was to use origin = <timestamp>, so something like origin = datetime.fromisoformat('1970-01-01 09:15:00+05:30') does the magic.

Overall code:

pd.DataFrame(download_data).set_index('date'['close']).resample('30T', origin=datetime.fromisoformat('1970-01-01 09:15:00+05:30')).ohlc()

Upvotes: 1

TFA
TFA

Reputation: 81

Though the answers provided here work as expected, it must be noted that both loffset and base are deprecated since version 1.1.0. The best and simplest way now is

pd.DataFrame(download_data).set_index('date'['close']).resample('30T', origin='start').ohlc()

This will set the start time to the first timestamp available in the dataframe.

Upvotes: 0

Jeril
Jeril

Reputation: 8521

There is one more way of doing it, you can use the base argument of resample:

pd.DataFrame(download_data).set_index('date'['close'].resample('30T', base=15).ohlc()

Upvotes: 3

jezrael
jezrael

Reputation: 862791

Solution is add parameter loffset in resample:

loffset : timedelta

Adjust the resampled time labels

df = (pd.DataFrame(download_data)
        .set_index('date')['close']
        .resample('30T', loffset='15min')
        .ohlc())
print (df)

                               open      high      low     close
date                                                            
2018-11-05 09:15:00+05:30  25638.25  25641.85  25589.3  25630.00
2018-11-05 09:45:00+05:30  25622.00  25745.00  25622.0  25714.85
2018-11-05 10:15:00+05:30  25720.05  25740.00  25692.9  25717.00
2018-11-05 10:45:00+05:30  25698.30  25744.75  25667.9  25673.95
2018-11-05 11:15:00+05:30  25680.30  25690.45  25642.9  25655.90

Upvotes: 1

Related Questions