search

lccasagrande / Deep-Knowledge-Tracing

An implementation of the Deep Knowledge Tracing (DKT) using Tensorflow 2.0
MIT License
94 stars 38 forks source link

some questions #2

Closed streamsteel closed 5 years ago

streamsteel commented 5 years ago

Your code is pretty good! But some troubles I have when I run the code:

lccasagrande commented 5 years ago

Hi streamsteel,

I read the paper where used 12 features to analyze the dataset and why you only group by user_id and skill_id?

I've grouped by user_id to form sequences of answers. The skill_id is encoded together with the answer because we want to find the probability of a student answering correctly a question B given its previous answers to other questions.

I want to carry out a CART (classify tree) base on the data.How to achieve？

I cannot help on this. I believe it's better to start a fresh project if you are not using Deep Learning and wont use keras to build your model.

how to use the model I saved where the file suffixed with .hdf5 to test data?

There is a jupyter notebook in this repository that you can follow to achieve this. Just omit the Part 4 as you are not going to train your model. The rest of the code should remain the same. Remember to put the path to your .hdf5 file in the "best_model_file" in Part 1.

streamsteel commented 5 years ago

Thank you very much for your answer. My current project needs to be implemented BKT(Bayesian knowledge tracking).Would you mind if I use part of your code to complete the paper?

lccasagrande commented 5 years ago

Not at all. Feel free to use it in your project, just not forget to give it the proper credits.