How to predict multivariate time series in a rolling manner?

[BEbillionaireUSD]

Hey,

I have a multivariate time series dataset looking like:

2017-05-25 08:36:00 30.86 0.0 1.45 161.56 46.46 0.0 ... 29.89 0.0 6.1250 3.85 144.95
2017-05-25 08:41:00 32.34 0.0 1.10 158.46 48.31 0.0 ... 29.46 0.0 2.9375 3.75 144.05
2017-05-25 08:46:00 31.45 0.0 1.05 160.93 47.15 0.0 ... 31.03 0.0 3.1250 4.15 138.88
2017-05-25 08:51:00 32.44 0.0 1.25 160.69 48.11 0.0 ... 30.11 0.0 3.0000 3.80 141.58
2017-05-25 08:56:00 30.75 0.0 1.20 151.66 50.21 0.6 ... 29.15 0.0 3.0000 3.85 146.77

with the format of DataFrame

I want to predict all these variates in a rolling way, i.e., given the 0-5 time slots' MTS, predict the 6-th slot's value (a multi-value vector); then, given the 1-6 time slots' MTS, predict the 7-th slot's value. How should I generate dataset? I don't know how to use MultivariateGrouper in this problem.

Thanks for any help in advance.

[satyrmipt]

@BEbillionaireUSD, do you managed to solve this task?