this is for holding javascript data
chengds edited The_technique_we_used_to__.tex
over 8 years ago
Commit id: bbbe01f8f6db0d574f6dedb4670e055c31e6a7c9
deletions | additions
diff --git a/The_technique_we_used_to__.tex b/The_technique_we_used_to__.tex
index 0dc00c4..79c01b2 100644
--- a/The_technique_we_used_to__.tex
+++ b/The_technique_we_used_to__.tex
...
The technique we used to turn a regression problem (of the type: estimate the correct level of a trait) into a classification problem is to separate the distribution into three parts: low values, middle values, high values. However, instead of having three classes, we focused on recognizing the low/high values and ignoring the middle ones. To separate the values we use the first and the third quartiles of each distribution. The following tables show the values of these quartiles and the number of users selected for the two resulting classes.
\begin{tabular}{c|cc|}
\hline \cline{2-3}
& \multicolumn{2}{|c|}{Ope} \\
\hline
Self \multicolumn{1}{|c|}{Self} & 2.0 & 4.0 \\
\hline
Attr \multicolumn{1}{|c|}{Attr} & -0.2 & 0.85 \\
\end{tabular}