Time Series in Python using Multiple Input Features

Hi All,
I have a Time Series problem of intraday prediction of ‘predicting work load on half hourly basis for each task id’. I have 300 task ids and daily half hour data for about 4 years.

Can some one please guide me what’s the best model that I can use for this problem in python?. This is a Univariate problem type and not multivariate as there is only 1 dependent(output) variable but am trying to predict for all tasks at a time for every half an hour.
I have checked ARIMA & SARIMAX models but I think they predict for 1 i/p feature at a time only that means I need to have 300 models individually but I want one model that can predict for 300 tasks in one go. I heard about VAR(Vector Auto Regression) and it works on multivariate problem ie, predicting multiple outputs. But not sure whether it can work for my problem. Can someone please give details on it and let me know how can I execute this using VAR or any other best model?
Ex:
Below is the sample snapshot of data that I have. date_timestamp is the time axis and I need to predict the values for each task id (I have total 300 task_ids) for each half hour from 6am to 11.30pm every work day.
|date_timestamp|task_id1|task_id2|task_id3|task_id4|task_id5|task_id6|…
|12/26/16 6:00|5|0|7|0|2|0|…
|12/26/16 6:30|2|0|4|3|0|9|…
|12/26/16 7:00|0|6|0|2|0|3|…
|12/26/16 7:30|1|0|4|0|5|2|…
|12/26/16 8:00|3|1|5|0|0|9|…
|12/26/16 8:30|0|6|0|0|1|0|…
|12/26/16 9:00|0|6|0|8|0|0|…
|12/26/16 9:30|7|0|1|3|0|5|…
|12/26/1610:00|6|3|5|0|8|2|…
Thanks a lot in adv!

© Copyright 2013-2019 Analytics Vidhya