Table 2.
F1score | AcMa | DaMa | AcFa | DaFa | |||||
---|---|---|---|---|---|---|---|---|---|
DKT | 0.74 | 0.09 | 0.72 | 0.0 | 0.79 | 0.0 | 0.79 | 0.0 | |
DKT+BN | 0.70 | 0.64 | 0.78 | 0.43 | 0.86 | 0.74 | 0.80 | 0.47 | |
F1score | AcCcf | DaCcf | AcA | DaA | Accuracy | ||||
DKT | 0.79 | 0.0 | 0.73 | 0.0 | 0.80 | 0.0 | 0.79 | 0.0 | 0.74 |
DKT+BN | 0.87 | 0.80 | 0.82 | 0.72 | 0.83 | 0.58 | 0.88 | 0.80 | 0.82 |
We repeated the experiments 20 times. The last column is the overall accuracy of the model. For each skill, the first column corresponds to the value of the f1score for predicting that the skill was incorrectly answered and the second column to the value of the f1score for predicting that the skill was correctly answered. The best ratio for each skill is in bold.