You can use the tabu package which offers you ready to use features to control the attributes (thickness, color) of the rules; alignment at the decimal separator can be achieved using the dcolumn package:
\documentclass{article}
\usepackage{xcolor}
\usepackage{dcolumn}
\usepackage{tabu}
\newcolumntype{A}{D{.}{.}{2.3}}
\begin{document}
\begin{table}[htbp]
\centering
\caption{Distribution of Number of Injured Non-Motorists by Injury Severity Level}
\extrarowsep=_3pt^3pt
\begin{tabu}to\linewidth{|[2pt gray]r|c|A|c|A|c|A|[1.5pt gray]}
\tabucline[1.5pt gray]-
\bfseries Injury Severity & \multicolumn{2}{c|}{\textbf{Pedestrian}} & \multicolumn{2}{c|}{\textbf{Bicyclist}} & \multicolumn{2}{c|[1.5pt gray]}{\textbf{All Non-Motorists}} \\
\tabucline[1.5pt gray]-
Possible injury & 1700 & 67.70\% & 502 & 59.40\% & 2202 & 65.60\% \\
Non-Incapacitating injury & 523 & 20.80\% & 259 & 30.70\% & 782 & 23.30\% \\\hline
Incapacitating injury & 250 & 10.00\% & 84 & 9.90\% & 334 & 9.90\% \\\hline
Fatal injury & 39 & 1.50\% & 0 & 0.00\% & 39 & 1.20\% \\
\tabucline[1.5pt gray]-
\bfseries Total & \bfseries 2512 & \multicolumn{1}{c|}{\bfseries 100.00\%} & \bfseries 845 & \multicolumn{1}{c|}{\bfseries 100.00\%} & \bfseries 3357 & \multicolumn{1}{c|[1.5pt gray]}{\bfseries 100.00\%} \\
\tabucline[1.5pt gray]-
\end{tabu}%
\label{tab:dvar}%
\end{table}%
\end{document}
Just for comparison purposes, the same table without the vertical rules and some of the horizontal ones:
\documentclass{article}
\usepackage{xcolor}
\usepackage{dcolumn}
\usepackage{tabu}
\newcolumntype{A}{D{.}{.}{2.3}}
\begin{document}
\begin{table}[htbp]
\centering
\caption{Distribution of Number of Injured Non-Motorists by Injury Severity Level}
\extrarowsep=_3pt^3pt
\begin{tabu}to\linewidth{rcAcAcA}
\tabucline[1.5pt gray]-
\bfseries Injury Severity & \multicolumn{2}{c}{\textbf{Pedestrian}} & \multicolumn{2}{c}{\textbf{Bicyclist}} & \multicolumn{2}{c}{\textbf{All Non-Motorists}} \\
\tabucline[1.5pt gray]-
Possible injury & 1700 & 67.70\% & 502 & 59.40\% & 2202 & 65.60\% \\
Non-Incapacitating injury & 523 & 20.80\% & 259 & 30.70\% & 782 & 23.30\% \\
Incapacitating injury & 250 & 10.00\% & 84 & 9.90\% & 334 & 9.90\% \\
Fatal injury & 39 & 1.50\% & 0 & 0.00\% & 39 & 1.20\% \\
\tabucline[1.5pt gray]-
\bfseries Total & \bfseries 2512 & \multicolumn{1}{c}{\bfseries 100.00\%} & \bfseries 845 & \multicolumn{1}{c}{\bfseries 100.00\%} & \bfseries 3357 & \multicolumn{1}{c}{\bfseries 100.00\%} \\
\tabucline[1.5pt gray]-
\end{tabu}%
\label{tab:dvar}%
\end{table}%
\end{document}
In my opinion, this second version looks much better, but I understand the original requirement for the thesis.
1. Put a name for the first column
There is already one, namely "File". Do you want to lower it to the second line? I wouldn't, because it is hierarchically on the same level as "Logistic..." and "Support...".
2. Center the words for "logistic reg" and "support vector machine"
Well, those are already centered, just not on the right columns. In your code, they are each centered on the first 2 columns below the \cline. Change the \multicolumn{2} to \multicolumn{3} and you'll probably get what I believe you mean.
3. Output with very small margin like a picture.
Check the geometry options if you are talking about page geometry, e.g.
\usepackage[landscape,scale=0.95]{geometry}
FYI, geometry enables you to change the geometry of a single page and resetting the usual geometry afterwards.
4. Other suggestions making the table looks professional.
Use the caption package to correct spacing for table captions: while table captions should be placed before the (tabular) content (as you did) and picture captions should be placed after the image, by default LaTeX puts vertical space before the caption label and not after it.
Use \toprule, \midrule, \cmidrule and \bottomrule from the booktabs package (that you are loading without using) instead of \hline and \cline.
Given that you are displaying figures meant to be compared, make the comparison easy to the eye: align your columns containing said figures right, in order to align the decimals. For even better results, use the dcolumn package as in jon's answer to automatically align the figures on their decimal and format them correctly (including using a minus sign instead of a hyphen). FYI, you can use the siunitx package in the rest of your document to format numbers (with \num) and units (with \si), but their use in tabulars is rather tedious. Also, for easy readability, units should go in your tabular's head, not inside the data itself.
Remove the unnecessary @{} and extra middle column; unless you know exactly what and why, (La)TeX does its typesetting pretty well on its own :)
Remove the \arraystretch / \ra "trick".
Given the "File" heading is structurally at the same level as the two "Regression" ones, use the same font for it. Also, put all your headings with a capital (or none, but be coherent).
Finally, to ease your own coding, I would suggest you format the content of your tabular with white spaces to visually separate the columns and resemble your output. I've found that editing tabulars made this way is much more easy, especially when you have to modify documents a few weeks / months after their creation.
Putting it all together
\documentclass{article}
\usepackage[landscape]{geometry}
\usepackage{booktabs,caption,dcolumn}
\newcolumntype{d}[1]{D{.}{.}{4}}% column type for figures with 4 decimals
\newcommand{\subhead}[1]{\multicolumn{1}{c}{#1}}% to format sub-headings of d-type columns
\begin{document}
\begin{table}
\centering
\caption{Caption}
\begin{tabular}{cd{4}d{4}d{4}d{4}d{4}d{4}}
\toprule
\textbf{File}& \multicolumn{3}{c}{\textbf{Logistic Regression}}& \multicolumn{3}{c}{\textbf{Support Vector Machine}}\\
\cmidrule(rl){2-4} \cmidrule(rl){5-7}
& \subhead{Sensitivity}& \subhead{Specificity}& \subhead{Total accuracy}& \subhead{Sensitivity}& \subhead{Specificity}& \subhead{Total accuracy}\\
\midrule
100 & 0.0790& 0.1692& 0.2945& 0.3670& 0.7187& 3.1815\\
101 & - 0.8651& 50.0476& 5.9384& - 9.0714& 297.0923& 46.2143\\
102 & 124.2756& - 50.9612& -14.2721& 128.2265& -630.5455& -381.0930\\
103 & 0.0357& 1.2473& 0.2119& 0.3593& - 0.2755& 2.1764\\
104 & - 17.9048& - 37.1111& 8.8591& - 30.7381& - 9.5952& - 3.0000\\
105 & 105.5518& 232.1160& -94.7351& 100.2497& 141.2778& -259.7326\\
\bottomrule
\end{tabular}
\end{table}
\end{document}
which yields:
Best Answer
Is this better?