How to calculate the Moodle's prediction models?

How to calculate the Moodle's prediction models?

by Motasem Al Haj Ali -
Number of replies: 3

Hi all, 


I am wondering if anyone know how to calculate the Moodle prediction models, as I am looking to compare them with some others has been created out side of Moodle. 

I know that I could decoded from the Moodle code but not sure, I post this subject here hoping that some one already worked on them before. 

Regards

Mo

Average of ratings: -
In reply to Motasem Al Haj Ali

Re: How to calculate the Moodle's prediction models?

by Dan Marsden -
Picture of Core developers Picture of Particularly helpful Moodlers Picture of Peer reviewers Picture of Plugin developers Picture of Plugins guardians Picture of Testers Picture of Translators
There's a wiki page here that describes it at a high level:
https://docs.moodle.org/en/Students_at_risk_of_dropping_out
Average of ratings: -
In reply to Dan Marsden

Re: How to calculate the Moodle's prediction models?

by Motasem Al Haj Ali -

Thanks Dan, 


I am looking for how the data has been extracted form Moodle's DB and what type of calculations the data has before feeding the machine learning algorithm with that data.  I was thinking to examine the models externally. 


Regards

Mo

Average of ratings: -
In reply to Motasem Al Haj Ali

Re: How to calculate the Moodle's prediction models?

by Dan Marsden -
Picture of Core developers Picture of Particularly helpful Moodlers Picture of Peer reviewers Picture of Plugin developers Picture of Plugins guardians Picture of Testers Picture of Translators
right - so first you need to read that doc linked above - and see how each activity is broken down into cognitive and social indicators - how those indicators are calculated is covered at a high level in that doc.

Each activity in moodle defines it's own indicator classes eg here's the social breadth one for the forum activity:
https://github.com/moodle/moodle/blob/master/mod/forum/classes/analytics/indicator/social_breadth.php

Which basically uses the community of inquiry class here:
https://github.com/moodle/moodle/blob/master/analytics/classes/local/indicator/community_of_inquiry_activity.php

The values calculated by Moodle for these indicators are stored in the mdl_analytics_indicator_calc table and then collected to be sent to the machine learning mode - to understand the indicator_calc table data you should also understand Moodle's "context" concept so that you can match the records in this table to a user.
https://docs.moodle.org/en/Context

Another good thing to be aware of is that when sampleorigin = "user_enrolments", the sampleid in the analytics_indicator_calc table relates to the id field in teh user_enrolments table (that's how you match to a user.)

hopefully that gets you a few steps closer to what you're looking for...
Average of ratings: Useful (2)