Determine well performing/bad performing individuals in a dataset considering a continuous quantitative variable I have a dataset with several individuals (let's say for example sellers). I have a very important information : the number of profit they generated. I also have several quantitative informations like age, wealth, experience with the job, rate of success when dealing with a client, days of rest during the year, etc...  
I want to know two things :
1- What are the variables which influence the most the profit generated by a seller
2- Considering this, which sellers performed well/poorly (for instance, let's say that rest and experience with the job are the most important variable, I want to tell that if a seller had a lot of rest and is inexperimented but generated 1000€ of profit, he is considered better than an experienced seller that generated 1500€)
How can I approach this problem ?
 A: For question 1:
It seems like a regression would answer your question, although I agree with the suggestion of PCA above.  You would get an estimate of the effects of each variable on profit:
In R, this would be something like (with your own field names):
model1<- lm(profit ~ age + experience + rest +...., data=dataname )
summary(model1)

You could see which values appear statistically significant, and assess whether you trust the model by testing the normal linear models assumptions (e.g. independence of residuals, equal variance of residuals etc.)  If you want to, refit the model omitting factors that are not significant (but recheck before using).
It is a bit more complicated if you have repeated measures, e.g. the same person measured each months.  These measurements will be correlated, and you need to adjust for this or you will get poor estimation of your model's standard errors.  If this is the case, you would be best looking at linear mixed model with random intercepts for clusters, such as lmer in the lme4 package in R, or maybe Generalised Estimating Equations (GEE) for the overall effects of each predictor.
For question 2
Use your model to predict expected profit and compare it to observed profit.  The ones with the highest ratio of observed profit/expected profit would be your 'best performers.'
