Data representation in a better way? I ran two programs A and B, and their corresponding features are a,b,c, and d. I would like to represent their features data in a better way instead of a table as shown below to explain why A is better than B or vice-versa. First, I thought about barplot representation, but feature a data is very small and that makes barplot representation as an improper one. Moreover, I also thought about barplot represnetatuon of A/B (normalization), but not satisfied with the resulting barplot graph. It would be great help if you share your ideas. I would like to use R for plotting the data.
prog  a   b    c    d
A    0.8 7900  70  27
B    0.3 1920 393  43

Here, the lower d values is better. Therefore, I would like to use the data of a,b, and c to explain why one program is better than other. In this case why A is better than B.
 A: The first question I would ask about these results is whether higher scores are always "better" for each of your four variables. If not, I would suggest that you would want at least two separate plots, one for variables where higher is better, and one where lower is better.
It's also not really good practice to place variables with different units on the same barplot. It's confusing at best and misleading at worst. Four separate barplots presented together, one for each variable, would seem like a fairly simple, clear solution. The only way I could see a single graph working is if you had baseline data for each group that allowed you to plot each score as a percentage increase/decrease.
A: You could take A (or B) as the reference software and express the values in %A.
Otherwise, if the four features represent something like speed, price, etc. you could use some symbol to represent them (e.g. 1 to 5 rabbits, 1 to 5 dollar signs etc, or just 1 symbol with area proportional to the value). Of course this is way less precise but it could be useful if this is for marketing purposes.  
A: What about icon plots, particularly Chernoff faces? They are apt to showing data composed of different units (and therefore magnitudes). Also, you could allocate your variables to features so adroitly that overall impression of face will convince that A is better than B despite that a, b, c indicate "betterness" positively but d indicate it inversely.
