python - ARIMA modelling for more than 1 time series -

so data looks this

           date1 date2 date 3.....date n instance1 instance2 . . . 

i don't want build arima model instance1. want universal model takes instances consideration. found lot of examples show me how fit for

         date1 date2 date 3.....date n instance1 

but none instances

if believe time-series correlated , want take correlations account in forecast/simulation, should @ vector auto-regression models (var). here couple options in python:

statsmodels

pyflux

if don't believe correlated, there's no reason can't loop through each time-series , apply arima model 1 @ time.