MAE, not MAD

nitroxides.tex

@@ -502,7 +502,7 @@ \subsection{Impact of the electrolytes} \label{sec:elect}
 \clearpage
 \subsection{Comparison to experiment} \label{sec:exp}
 
-A comparison between theoretical (including all corrections discussed above) and experimental oxidation potentials is shown in Fig.~\ref{fig:expvstheo}.\todo{This is not really $E^f$ for the moment ;)} Excluding compounds bearing an \ce{NH2} group (\textbf{51}, \textbf{57}, and \textbf{59}) results in an excellent linear correlation ($R^2>0.9$). The computed potentials for these compounds are higher (by $>\SI{0.1}{\volt}$) than the fit line, suggesting that the amine group might be protonated under the experimental measurement conditions. This is indeed consistent with other cases (\textit{e.g.}, \textbf{25} vs \textbf{35}) discussed in Section \ref{sec:sar}. The remaining results indicate that the method used in this paper tends to overestimate the oxidation potentials in water and underestimate them in acetonitrile, with a large mean absolute deviation (MAE).
+A comparison between theoretical (including all corrections discussed above) and experimental oxidation potentials is shown in Fig.~\ref{fig:expvstheo}.\todo{This is not really $E^f$ for the moment ;)} Excluding compounds bearing an \ce{NH2} group (\textbf{51}, \textbf{57}, and \textbf{59}) results in an excellent linear correlation ($R^2>0.9$). The computed potentials for these compounds are higher (by $>\SI{0.1}{\volt}$) than the fit line, suggesting that the amine group might be protonated under the experimental measurement conditions. This is indeed consistent with other cases (\textit{e.g.}, \textbf{25} vs \textbf{35}) discussed in Section \ref{sec:sar}. The remaining results indicate that the method used in this paper tends to overestimate the oxidation potentials in water and underestimate them in acetonitrile, with a large mean average error (MAE).
 
 
 \begin{figure}[!h]