Closed josephtey closed 7 years ago
Hi, Joseph
Thanks for using our toolbox.
The current design of the algorithm does not allow for multiple interactions. Our solution is to use the latest attempt in the sequence. This is not optimal but we have no immediate plan to extend it.
In addition, recently, it comes to our attention that pyirt does not pass unittest in python 3.X. The estimated parameters shrink toward ZERO. We expect to release an update to fix the bug.
Jc
Hi, Joseph. We just released a version that converges qualitatively. Although it still does not allow for multiple attempts, the code at least gives you correct results in a large sample.
Hi there, Very good work; this is definitely the easiest IRT library existing for python. I'm using this library to determine item and student params for vocabulary learning - that is, a student will often revisit the same word over and over again.
Is this library designed for multiple student-item interactions? Or is it solely for a test-based assessment?
Thanks