Skip to content
Browse files

two talks for EDM memphis, 2013

  • Loading branch information...
1 parent b4c380c commit 96a75d5e26516c3d783b389203138a3ddbc3d7c1 @bvds committed Jul 6, 2013
View
3 LogProcessing/moment-of-learning/bayesian-knowledge-tracing.tex
@@ -58,7 +58,8 @@
\linespread{1.25}\selectfont
-\def\runningfoot{}
+\def\runningfoot{\thepage \fill Journal of Educational Data Mining, Volume 5, Issue 2
+(in press)}
\def\firstfoot{}
\begin{document}
View
4,180 LogProcessing/moment-of-learning/fit-models.nb
3,610 additions, 570 deletions not shown because the diff is too large. Please use a local Git client to view these changes.
View
BIN LogProcessing/moment-of-learning/memphis-moment-of-learning.odp
Binary file not shown.
View
BIN LogProcessing/moment-of-learning/memphis-three-models.odp
Binary file not shown.
View
10 LogProcessing/moment-of-learning/three-models.tex
@@ -33,8 +33,8 @@
data with the goal of seeing which model best describes individual
student learning of a particular skill. The log data is from students
who used the Andes intelligent tutor system for an entire semester of
-introductory physics. We discover that, in this context, the "best
-fitting model" is not necessarily the "correct model" in the usual
+introductory physics. We discover that, in this context, the ``best
+fitting model'' is not necessarily the ``correct model'' in the usual
sense.
\end{abstract}
@@ -585,7 +585,7 @@ \subsection{Conclusions}
particular model $\mathcal{A}$ (for some choice of model parameters),
then we can safely conclude that model $\mathcal{A}$ is the correct model.
-In other words, when we fit individual student data to a model (fitting
+When we fit individual student data to a model (fitting
model parameters separately for each student), then we can make no
statements about what model is ``correct'' in the sense that it may
have generated the data. We can only talk about a model being a good
@@ -644,8 +644,8 @@ \subsection{Conclusions}
Finally, we see that the scatter plot of Akaike weights for student
data is remarkably similar to the scatter plots for the random model.
This suggests that the student data has a high degree of randomness,
-and, in general, that study of the random model may be quite useful for better
-understanding the student data.
+and, in general, that the random model may be quite useful as a
+reference point or ``control condition'' when trying to understand the student data.
%ACKNOWLEDGMENTS are optional
\section{Acknowledgments}

0 comments on commit 96a75d5

Please sign in to comment.
Something went wrong with that request. Please try again.