Wrapper around liblinear-tools

For classification, only L2-regularized logistic regression is supported. For regression, only linear SVR. When doing classification with bagging, each model is trained on balanced bootstraps from the training set (one bootstrap for the positive class, one for the negative class). The size of the bootstrap is the size of the smallest (under-represented) class.

usage: linwrap -i <filename>: training set or DB to screen output file C epsilon (for SVR); (0 <= epsilon <= max_i(|y_i|)) w1 gnuplot of bags for bagging (default=off) of cross validation scan for a trained model (requires n>1) also requires (c, w, k) to be known random seed set portion (in [0.0:1.0]) from .AP files (atom pairs; will offset feat. indexes by 1) set (overrides -p) set (overrides -p) set (overrides -p) mode; use trained models mode; save trained models overwriting existing model file for best C scan #steps for SVR ; also, implied by -e and --scan-e weight to counter class imbalance range for w (semantic=start:nsteps:stop) [--c-range <float,float,...>] explicit scan range for C (example='0.01,0.02,0.03') [--k-range <int,int,...>] explicit scan range for k (example='1,2,3,5,10') number of bags (advice: optim. k rather than w)