yearprediction.ssc

:silent
import BIDMach.models.RandomForest

class xopts extends Learner.Options with MatDS.Opts with RandomForest.Opts with Batch.Opts
val opts = new xopts

val mdir = "/code/BIDMach/data/YearPrediction/"
val data = loadFMat(mdir+"train.fmat.lz4");
val cats = loadFMat(mdir+"cats.fmat.lz4");

val rr = rand(1,data.ncols);
val itrain = find(rr > 0.1f);
val itest = find(rr <= 0.1f);
    
val train = data(?,itrain);
val traincats = cats(?,itrain);
val test = data(?,itest);
val testcats = cats(?,itest);
val preds = zeros(testcats.nrows, testcats.ncols);

val ds = new MatDS(Array(train.asInstanceOf[Mat],traincats), opts);

val dsp = new MatDS(Array(test.asInstanceOf[Mat],preds), opts);

val nn = new Learner(             // make a learner instance
    ds,                           // datasource
    new RandomForest(opts),       // the model 
    null,                         // list of mixins or regularizers
    new Batch(opts),              // the optimization class to use
    opts)                         // pass the options to the learner as well

val mm = new Learner(             // make a predictor
    dsp,                          // datasource
    nn.model,
    null,
    null,
    opts)                         // pass the options to the learner as well


opts.useGPU = true
opts.batchSize = 20000
opts.depth = 20
opts.ntrees = 25
opts.ncats = 90

opts.nsamps = 25
opts.nnodes = 250000
opts.nbits = 16
opts.gain = 0.001f
opts.regression = true
opts.autoReset = false

val rf = nn.model.asInstanceOf[RandomForest]

nn.train

opts.training=false
opts.putBack = 1

mm.predict

val rr=zeros(nn.results.ncols/opts.depth, opts.depth);
rr(?) = nn.results(0,?).t
val diffs = DMat(preds - testcats);
:silent
val scores = mean(rr)
val MSE = (diffs dotr diffs) / diffs.length;