Titus' Lab Notebook
This is where I will write up the list of experiments that I am
planning on performing while working toward my goal of being able to
tell (hopefully from some combination of question text and per-item
scores) which questions in a course are most related, the topics of
those groupings, and the proficiencies of every student in the course
with respect to those topics.
To Do
- Come up with a straw-man for my question difficulty
measurements.
- Check for a more intuitive measurement for instrument
effectiveness
- Scale cs164mtweighted by adding duplicate samples instead of
scaling the vectors
- PCA on stemmed question text for clustering.
Experiments