How to train the model

I have 2 collections, one of them contains the static data, “students”, the other one (“marks” ) contains the user_id, average_mark and date.

For every quarter I need to go and look at the data and write a short overview.

For example: “Dave has shown remarkable improvement in the first quarter of 2024, with his average mark increasing from 8.2 to 8.7”.

My collections are:

students: [
     {_id: 1, name: "Dave"}
marks: [
     {_id: 1, student_id: 1, average_mark: 8.2, date: ISODate("2023-09-30T00:00:00.000+0000")},
     {_id: 2, student_id: 1, average_mark: 8.7, date: ISODate("2023-12-31T00:00:00.000+0000")},

Is it possible for me to train the model based on the past data, where I would feed the raw data and my (manually) written overview? So that in the future I can input the data and generate the text?