Additive factors model and additive factors model with slip implemented in python
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.gitignore
LICENSE.txt
README.md
afms_workflow_predict.py
bounded_logistic.py
custom_logistic.py
models.py
plot_datashop.py
process_datashop.py
roll_up.py
util.py

README.md

AFM and AFM+S in Python

This is a python implementation of the Additive Factors Model (Cenn, 2009) and The Additive Factors Model with Slipping Paramters (MacLellan et al., 2015). Wherever possible I tried to maintain scikit-learn convention, so that the code would be compatible with their helper functions (e.g., for cross validation).

custom_logistic.py is an estimator that can be used to implement AFM. bounded_logistic.py is an estimator that can be used to implement AFM+S.

process_datashop.py can be used to read a student step or transaction export from DataShop and to run AFM and AFM+S on it. Alternatively, the data can be passed to the scripts as long as it adheres to the tab-delimited DataShop format.This script can be called with the -h argument to get details on supported arguments. By default the script runs AFM+S when passed a student step export from Datashop; e.g., $ python process_datashop.py student_step_datashop_export.txt

plot_datashop.py can be used to plot student learning curves with the learning curves predicted by AFM and AFM+S. Similar to process_datashop.py, it accepts student step files in datashop format; e.g., $ python plot_datashop.py student_step_datashop_export.txt

For a slightly more detailed example of how to use these scripts see my blog post.

Citing this Software

If you use this software in a scientific publiction, then we would appreciate citation of the following paper:

MacLellan, C.J., Liu, R., Koedinger, K.R. (2015) Accounting for Slipping and Other False Negatives in Logistic Models of Student Learning. In O.C. Santos et al. (Eds.), Proceedings of the 8th International Conference on Educational Data Mining. Madrid, Spain: International Educational Data Mining Society (pdf).

Bibtex entry:

@inproceedings{afmslip:2015,
author={MacLellan, C.J. and Liu, R. and Koedinger, K.R.},
title={Accounting for Slipping and Other False Negatives in Logistic Models
of Student Learning.},
booktitle={Proceedings of the 8th International Conference on Educational
Data Mining},
editor={Santos, O.C. and Boticario, J.G. and Romero, C. and Pechenizkiy, M.
and Merceron, A. and Mitros, P. and Luna, J.M. and Mihaescu, C. and Moreno,
P. and Hershkovitz, A. and Ventura S. and Desmarais, M.},
year={2015},
publisher={Interational Educational Data Mining Society},
address={Madrid, Spain}
}