I have a Panel data set as below except with a lot more variables. Can I perform a classification to understand the chances that a student will leave the college ? I want to be able to classify the student with a probability that he/she will leave the college.

The dependent variable is a binary variable Yes / No. Yes - Student has left the college without completion No - Student is still enrolled Graduated - This is another category which could be considered as student still with the college.

Some variables change over time, some don't. For example age of the student, economic status (student can move between localities), GPA.

Also some students commence course in 2016, some in 2017. So a student may have 3, 2 or 1 row, if I consider 3 years of data.

I was told to drop the Student ID column and treat the rows as independent of each other and run classification models like Random Forest. Would that be correct ?

Socio Economic Indicator is based on the locality a student lives at the time of data collection. If he or she moved to a different locality the following year, the Socio Economic Status may change.

If I have 1 row per student I could run Random Forest and oter classification algorithms. If I have panel data I understand Fixed Effects and Random Effects are he model used. Now, this suggestion of dropping student ID's and running classification models, doesn't seem right to me.

enter image description here



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.