siyuanzhao/2016-edm

Source code and data sets for Going Deeper with Deep Knowledge Tracing (EDM-2016)

2016-edm's Issues

better result?

hello, i just run you project without any changes, and i get a better result with AUC up to 0.815, compare to 0.75 in you paper on dataset 09_10(C). i want to know why, thank you?

CAT数据和paper里面统计的数据不一致

skill的数目只有99， record也只有500000出头

About data

Can you explain what is represented by each line in the data? I am a bit confused, what is the difference between indexID and problemID in Table4?
Can you help me?Thank you very much.

这里只有DKT的实现，能否提供一下PFA的实现？

我看到这篇论文Table 2. An example of PFA’s training data里面列出的训练数据还有一项diffculty。我不太理解！经典的Performance Factor Analysis里面，参数beta代表知识点的难易度。但是beta不是训练输入的数据，而是逻辑回归要拟合的参数。是计算结果，不是计算前预估吧？你文章里提到的diffculty是代表什么？

bug

when it is "writing hyperparameters into file",then Segmentation fault (core dumped) comes.What i do to solve it?

What is the difference between Data (b) & (c)

I wonder what is the difference between Data 09-10 (b) & (c).

In paper, Repeat and Combine is difference.

But when I see the table, I can't find the that meaning.

If the problem's skill is combine of 128 & 140.
I understand (b) case
input is such like

128 140
1 1

I think it is repeat sequence.

And what's the difference 09-10(c) ???

Please help self-studying student...

When Second learning time, this error is comming

1. first run
   tensorflow/core/framework/op_kernel.cc:940] Invalid argument: indices[1635] = [1635,230] is out of bounds: need 0 <= index < [4818,222]
2. second run
   tensorflow/core/framework/op_kernel.cc:940] Invalid argument: indices[1] = [1,231] is out of bounds: need 0 <= index < [4818,222]

I wonder why this situation is made.

Something wrong in input tensor size

Size of the input is changed randomly,(1635, 1)

Please check about this issue.