Language eexperiments.

paper/experiments.tex

@@ -305,7 +305,7 @@ It is able to learn model parameters that capture the behavior of decision maker
 %\contraction shows consistently poorer performance, but not dramatically worse.
 % THE VARIATION IS DRAMATICALLY WORSE
 %
-Again, our interpretation is that this is due to the fact that \contraction crucially depends on the data points that correspond to the most lenient decision makers, while \cfbi makes full use of all data.
+Again, our interpretation is that this is due to the fact that \contraction crucially depends on the cases assigned to the most lenient decision makers, while \cfbi makes full use of all data.
 
 
 
@@ -353,7 +353,7 @@ In these settings, the decisions in the data are made mostly based on background
 % WHAT??? NOT AS GOOD
 Nevertheless, the proposed method (\cfbi) is able to evaluate different decision makers $\machine$ accurately. 
 %
-\contraction shows again consistently worse performance in comparison, In comparison to \contraction in the basic case (Figure~\ref{fig:results_errors}) the performance is also worse, indicating some sensitivity to unobservables.
+\contraction shows again consistently worse performance in comparison. Furthermore, when compared to the basic case (Figure~\ref{fig:results_errors}), the performance of \contraction is also worse, indicating some sensitivity to unobservables.
 
 % SOUNDS AGAIN THAT WE MADE THE FIGURE FOR THIS
 % AND THE READER WONT UNDERSTAND
@@ -363,6 +363,13 @@ Nevertheless, the proposed method (\cfbi) is able to evaluate different decision
 
 Thus overall, in these synthetic settings our method achieves more accurate results with considerably less variation than \contraction, allowing for evaluation in situations where the strong assumptions of contraction inhibit evaluation altogether.
 
+\begin{figure}
+\begin{center}\includegraphics[width=\linewidth,trim={0 0 0 0.25cm},clip]{img/sl_errors_betaZ5}
+\end{center}
+\caption{Error of estimate w.r.t true evaluation when the effect of the unobserved $\unobservable$ is high ($b_\unobservable=5$). Although the decision maker quality is poorer, the proposed approach (\cfbi) can still evaluate the decision accurately. \contraction shows higher variance and less accuracy.}
+\label{fig:highz}
+\end{figure} % RL: Note that only machine decision maker is poorer, not the human.
+%\subsection{Results}
 
 \subsection{COMPAS data}
 
@@ -414,13 +421,7 @@ The built logistic regression model was used in decision maker \machine in the t
 The deployed machine decision maker was defined to release \leniencyValue fraction of the defendants with the lowest probability for negative outcome.
 
 
-\begin{figure}
-\begin{center}\includegraphics[width=\linewidth,trim={0 0 0 0.25cm},clip]{img/sl_errors_betaZ5}
-\end{center}
-\caption{Error of estimate w.r.t true evaluation when the effect of the unobserved $\unobservable$ is high ($b_\unobservable=5$). Although the decision maker quality is poorer, the proposed approach (\cfbi) can still evaluate the decision accurately. \contraction shows higher variance and less accuracy.}
-\label{fig:highz}
-\end{figure} % RL: Note that only machine decision maker is poorer, not the human.
-%\subsection{Results}
+
 
 \begin{figure}
 %\centering