Cfrgtkky

I have a pandas dataframe with columns run, type, vectime and vecvalue.
For rows (probably not the correct term here) with type=="vector" the vectime and vecvalue contains arrays:

loaded_csv.tail(2)



     run    type    module  name    attrname    attrvalue   value   vectime vecvalue

709 run-id1 vector  client499.app[0]    packetReceived:vector(packetBytes)  NaN None    NaN [20.02926688, 20.04433504, 20.04556544, 20.059...   [1460.0, 1460.0, 1460.0, 1460.0, 1460.0, 1460....

710 run-id1 attr    client499.app[0]    packetReceived:vector(packetBytes)  checkSignals    False   NaN None    None

The 'vectime' contains time points for the values in 'vecvalue'.

I have manged to run simple things such as apply min() max() to the array and assigning the value to the original dataframe by running:

loaded_csv.loc[(loaded_csv.type == "vector"),('min_vectime')] = loaded_csv.loc[(loaded_csv.type == "vector"),('vectime')].apply(min)

Which in my understanding 'filters' rows with type == vector and adds new column 'min_vectime' with the appropriate value.

However I am unable to figure out how to do similar thing with resample on 'vecvalue' using 'vectime' as index and replacing the original values (with the resampled ones) or adding new columns.

I found a way how to achieve part of it by getting the data from the dataframe, create Series and apply the resample.

for row in loaded_csv[loaded_csv.type == "vector"].itertuples():

    print_series = pd.Series(row.vecvalue, index = pd.to_timedelta(row.vectime,  unit='s'))

    result = print_series.resample('10ms',label='right').sum()*8

    resultPlot = result.rolling( window=30, min_periods=10).mean()

I am looking for a cleaner and less complicated way.

There are many examples on each of these tasks - slicing dataframe, applying function whether using .resample directly or via .apply() but on a simpler/different dataframe structure and I get lost somewhere in between.

Partial solution

I found a way how to turn the arrays into Series and resample it but when I try to insert back to the dataframe as a new column I get only NaN.
I had wrap it in pd.Dataframe() to get it included in the resulting Dataframe.
But now I have a column with Dataframe [Series] instead of just Series.

loaded_csv.loc[(loaded_csv.type == "vector"),('resampled')] = loaded_csv.loc[(loaded_csv.type == "vector"),('vectime', 'vecvalue')].apply(lambda x:  pd.DataFrame(pd.Series(x.vecvalue, index = (pd.to_timedelta(x.vectime, unit='s')))), axis= 1)

loaded_csv.loc[(loaded_csv3.type == "vector"),('resampled')].head()



76                           0

00:00:20.029266  1460....

157                          0

Name: resampled, dtype: object