Question: unequally spaced timeseries about adtk HOT 5 CLOSED

try s_train.resample('15min').ffill() if adtk.data doesn't have resample

from adtk.

@ajdapretnar Seasonal decomposition in ADTK requires the input follows equally spaced time series, as you noticed. Therefore, the time series should be resampled with a constant frequency, for example 15 min.

If you got a ValueError regarding NaN values, the reason is that your resampling may introduce NaN value to the new time series (in your example, 9:45 will have NaN value). Currently, SeasonAD does not support time series with NaN, unless the NaN values are on the starting or ending part of the time series and they will be ignored.

In the adtk.data module, we offer a resample function, which resamples a time series with user-given space. The resampling is based on linear interpolation. You may try it like the follows.

from adtk.data import resample
s_train = resample(s_train, dT="15 min")

If you want to fill NaN with forward or backward filling instead of interpolation, I believe you can also use the fillna method of Pandas.

from adtk.

Thanks, this is useful! Will try it.

from adtk.

My df is not having any null values but i am still getting this error when going to higher samples.
I have minute by minute data and when i am going above 7days i am getting this error. How many data points it can handle because when i am doing this for 10 days above its not givign any anomalies.

from adtk.

Removed adtk.data.resample because its functionality is highly overlapped with pandas resampler module

from adtk.