The graphical package TikZ can be used in easy way to provide fancy looking table with improved readability. The following commands produce a box, which fill corresponds to a median value. Then confidence intervals are added to the median value. The box spans one row and the actual numbers span the adjacent row.
Commands to draw a box:
\newcommand{\drawBox}[3]
{
\begin{tikzpicture}
\def\w{1.5} % width of a box
\def\x{#1/100*\w} % median value
\def\xl{#2/100*\w} % lower confidance interval
\def\xu{#3/100*\w} % upper confidance interval
% draw rectangle (black border)
\filldraw[fill=gray!#1!white!, draw=black] (0,0) rectangle (\x,0.2);
\draw [gray] (0,0) rectangle (\w,0.2); % fill with the grey
% draw upper confidance interval
\draw (\x,0.1) -- (\xu,0.1) -- (\xu,0.15) -- (\xu,0.05);
% draw lower confidance interval
\draw (\x,0.1) -- (\xl,0.1) -- (\xl,0.15) -- (\xl,0.05);
\end{tikzpicture}
}The interface command for tables:
\newcommand{\boxNumberConf}[3]
% NOTICE the & for a column separation
{\drawBox{#1}{#2}{#3} & #1 (#2--#3)} Example of use:
\boxNumberConf{50}{45}{55} % {median}{lower CI}{upper CI}Example of results presentation in a Table:
\begin{table}[ht!]\small
\centering
\caption{\label{tab:classification_feature_sel} Classification results for
different groups. The results are average across all folds of CV (50x4 folds CV) and presented using median (25th -- 75th) percentiles.}
\begin{tabular}{p{1.5cm} r lr lr lr}
\toprule
Feature set&[\%]& \multicolumn{2}{c}{NaiveBayes} & \multicolumn{2}{c}{SVM} & \multicolumn{2}{c}{C4.5 Tree}\\
\midrule
\multirow{4}{1.5cm}{HRV-based}
&SE & \boxNumberConf{53}{47}{63} & \boxNumberConf{53}{44}{60} & \boxNumberConf{47}{40}{60}\\
&SP & \boxNumberConf{74}{71}{77} & \boxNumberConf{76}{72}{79} & \boxNumberConf{75}{70}{79}\\
&PR & \boxNumberConf{21}{18}{24} & \boxNumberConf{21}{18}{24} & \boxNumberConf{19}{16}{23}\\
&F & \boxNumberConf{30}{27}{34} & \boxNumberConf{29}{25}{34} & \boxNumberConf{28}{23}{32}\\
\midrule
\multirow{4}{1.5cm}{Complete set}
&SE & \boxNumberConf{60}{53}{67} & \boxNumberConf{53}{47}{60} & \boxNumberConf{38}{27}{47}\\
&SP & \boxNumberConf{75}{72}{77} & \boxNumberConf{78}{75}{80} & \boxNumberConf{81}{75}{85}\\
&PR & \boxNumberConf{23}{20}{25} & \boxNumberConf{23}{20}{26} & \boxNumberConf{19}{15}{23}\\
&F & \boxNumberConf{33}{29}{36} & \boxNumberConf{33}{28}{37} & \boxNumberConf{25}{19}{31}\\
\bottomrule
\end{tabular}
\end{table} \documentclass[a4paper,11pt,oneside]{report}
\usepackage[english]{babel}
\usepackage{tikz}
\newcommand{\drawBox}[3] % drawing the bog
{
\begin{tikzpicture}
\def\w{1.5} % width of a box
\def\x{#1/100*\w} % median value
\def\xl{#2/100*\w} % lower confidance interval
\def\xu{#3/100*\w} % upper confidance interval
\filldraw[fill=gray!#1!white!, draw=black] (0,0) rectangle (\x,0.2);
\draw [gray] (0,0) rectangle (\w,0.2); % fill with the grey
\draw (\x,0.1) -- (\xu,0.1) -- (\xu,0.15) -- (\xu,0.05);
\draw (\x,0.1) -- (\xl,0.1) -- (\xl,0.15) -- (\xl,0.05);
\end{tikzpicture}
}
\newcommand{\boxNumberConf}[3] % interface command
{\drawBox{#1}{#2}{#3} & #1 (#2--#3)} % NOTICE the & for col. sep.
%%%%%%%%%%%%%%%%%%%%%%%%
\begin{document}
\begin{tabular}{l r lr}
& \multicolumn{2}{c}{mean (95\% CI)} \\
Sensitivity & \boxNumberConf{74}{65}{81} \\
Specificity & \boxNumberConf{90}{81}{95} \\
\end{tabular}
\end{document}Results of the minimal example:

