The implementation of the paper Graph-based Knowledge Tracing: Modeling Student Proficiency Using Graph Neural Network.
The architecture of the GKT is as follows:
To run this code you need the following:
- a machine with GPUs
- python3
- numpy, pandas, scipy, scikit-learn and torch packages:
pip3 install numpy==1.17.4 pandas==1.1.2 scipy==1.5.2 scikit-learn==0.23.2 torch==1.4.0
Note that don't use pandas with 0.23.4 version, because it will cause bugs when perform the following command in the processing.py file.
df.groupby('user_id', axis=0).apply(get_data)
If you use 'assistment_test15.csv' file to test, then in pandas 0.23.4 version, after groupby users, it will return 16 students. But if you use pandas in 1.x version, it will return 15 students. (This bug is found by vinnnan)
Use the train.py script to train the model. To train the GKT model on ASSISTments2009-2010 skill-builder dataset, simply use:
python3 train.py --data-file=skill_builder_data.csv --model=GKT --graph-type=Dense
We also provide the baseline, i.e. Deep Knowledge Tracing(DKT) for performance comparison. To train the DKT model on ASSISTments2009-2010 skill-builder dataset, simply use:
python3 train.py --data-file=skill_builder_data.csv --model=DKT
You might want to at least change the --data_dir and --save_dir which point to paths on your system to save the knowledge tracing data, and where to save the checkpoints.
