The importance of a good education system for societal growth has been recognized widely. This work addresses the problem of optimally grouping students in a class so that the overall gain due to cooperative learning summed over all students is maximized. We consider the setting in which every student can learn from his peers and work with two functions for learning gain. In one, a student is able to increase his score to the average score of higher ability peers. In the other, he potentially increases his score to their median score. We present fast partitioning algorithms for both the cases.
Algorithms
- Mean_Partitions: Partitioning to maximize learning gain, where students can improve up to mean of higher ability peers.
- Median_Partitions: Partitioning to maximize learning gain, where students can improve up to median of higher ability peers.
Code
The source codes used in the paper can be obtained from here.
Datasets
- GATE 2016 Scores: Scores of candidates taking Graduate Aptitude Test in Engineering (GATE) for year 2016 in various disciplines.
- SSC CGL 2016 Scores: Scores of candidates from different regions of India in SSC Combined Graduate Level (CGL) Exam for year 2016. The raw data used, is publicly available from http://ssc.nic.in.
- Stack Exchange Up-Votes: Number of Up-Votes acheived by users on different topics over Stack Exchange platform. The raw data used, is publicly avialble from https://archive.org/details/stackexchange.
The processed datasets used in experiments can be obtained from here.
Bibtex Entry
@techreport{agrawal2017optimal,
title={Optimal Partitioning of Students for Maximizing Learning Gain},
author={Agrawal, Rakesh and Nandanwar, Sharad and Murty, M. N.},
year = {2017},
month = {February},
number = {TR-2017-001},
institution = {Data Insights Laboratories},
address = {San Jose, CA 95120}
}