the error of running tutorial example kats_101_basics.ipynb about kats HOT 3 CLOSED

Hello @jeffhandl, @wenouyang

I was able to reproduce the issue on windows and it turns out that it's a known issue when attempting to convert a datetime to a negative timestamp (before 1970).

Here is a stack overflow post regarding the error on windows
https://stackoverflow.com/questions/56931738/python-crash-on-windows-with-a-datetime-close-to-the-epoch?noredirect=1#comment100413591_56931738

and a link to the issue in python itself
https://bugs.python.org/issue37527

This problem can be currently be sidestepped on windows by not attempting the unix timestamp conversion and using other methods instead. Other methods are outlined inside of the kats_101_basics.ipynb in the surrounding cells.

Some of the alternatives are:

# Passing the entire dataframe to the TimeSeriesData Object
air_passengers_ts = TimeSeriesData(air_passengers_df)

# Passing the series of string dates
air_passengers_ts_from_series = TimeSeriesData(time=air_passengers_df.time, value=air_passengers_df.value)

# Passing in a series of python datetime objects in a series 
from dateutil import parser

air_passengers_ts_datetime = air_passengers_df.time.apply(
    lambda x: parser.parse(x)
)

ts_from_unixtime = TimeSeriesData(
        time=air_passengers_ts_datetime , 
        value=air_passengers_df.value, 
        use_unix_time=True, 
        unix_time_units="s"
)

from kats.

Hi @wenouyang, thank you for submitting this issue. I'm not able to reproduce it and am wondering if it's related to the version of Pandas you are using and this known issue. Does this error persist after upgrading Pandas?

from kats.

@Axemen , @jeffhandl , thank you very much for your response. The issue has been solved as you suggested.

from kats.