search

Breakend / experiment-impact-tracker

MIT License
273 stars 31 forks source link

Nan values and pandas #53

Open pnovoa opened 3 years ago

pnovoa commented 3 years ago

Hi again, could you please take a look at the aggregation functions (e.g. mean) that you use with pandas Dataframe objects? Occasionally, a "TypeError: Could not convert .... to numeric" occurs. Apparently this error is caused due to the presence of some "nan" values in the column of the dataframe to be summarized. Maybe a solution would be to invoke the mean function as follows: dataframe.mean (skipna = True). Thank you!

pnovoa commented 3 years ago

I experienced the aforementioned problem again and tried to fix it with the following code in powercap.py (_parse function):

dt = pd.to_numeric(data[col], errors="coerce") summary[col] = dt.mean()

Breakend commented 3 years ago

Hi, can you provide us with a stack trace for the error by any chance? We'd like to reproduce this. Also what platform are you running this on?