A very short term predicting sequences system

Job ID: 32359685

Budget: €30 – €250 EUR

On a system alert, (LINUX ) which provides every 5 minutes, a sequence of 12 alerts numbers, there is a need to predict at least 7 alerts numbers for the next row.
This is a very Short term memory system, because, I am able to guess 7 alerts from just 10 previous input datalines. and able to perform corrective actions.

Very often, there are some combinations of alert numbers which are repeting at every row, or every two rows. because alerts are connected

This script in PYTHON (rel >= V3.6 )should train on all previous datalines, except the last one for testing. it's more a probabilistic system . May be no training or testing is required
, not sure there are some patterns. Alerts are too aleatory

May be, an LSTM could be the answer

please, propose some ideas to solve this