Using Adaboost for binary classification

Hi Guys - Have another question regarding AdaBoost algorithm. I have a binary classification problem with 20 predictor variables and 2 classifications (1 & -1). I know that that the predictor variables are non-linear so I would like split each one into (say) quartiles.

I still want the final outcome to produce a coefficient at the predictor level. So ultimately I would have an input value consisting of the predictor variable and it's quartile, and I would use both to estimate the classification.

Can you advise how I would go about this?

Thanks for help Dan