When I run the following:
I get a dataframe with dates converted to epoch.
It happens on line 253 in time_series.py:
df = pd.DataFrame(dict(Date=timestamp_column, Security=symbol_column,
This doesn't always happen, but when it does it seems obviously related to forcing everything to float, and then the failure of df.convert_dtypes a couple lines below.
Also there is the future warning about that approach, which notes that if all values in the dataframe can't be converted to float, you'll get a TypeError.
Any suggestions (aside from not using normalize, which is the only reason that code is running) ?
@dmears I can replicate your issue if I am using an older version of the Eikon library -
However - if I try in the codebook app - I get the correct dates with normalize=True
I don't know what version of the Eikon Lib you are running but if you install the latest refinitiv.dataplatform library (pip install refinitiv.dataplatform) it should work as expected:
import refinitiv.dataplatform.eikon as ek ek.set_app_key('YOUR APPKEY HERE') test=ek.get_timeseries(['.SPXTR','.TRSPTTEN'],start_date='2021-11-01',end_date='2021-11-23',fields=['CLOSE'],interval='daily',calendar='tradingdays',corax='adjusted',normalize=True) test
I hope this can help.
Thanks for the reply!
I wasn't aware of dataplatform, I've been using conda, and the latest version of eikon on conda-forge is 1.1.10
i tried using pip to install eikon 1.1.14 and dataplatform as well, in a new conda env, but still getting the same error.
I have the following packages in the new env:
@dmears I checked codebook and it is using pandas 1.1.5 - from your package list I would downgrade httpx (pip install httpx==0.19.0) and nest-asyncio too (pip install nest-asynchio==1.3.3) - also what version of python are you running - it seems maybe 3.9x - I would drop to 3.8.8 and try again. I hope this can help.