lars.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from modshogun import RealFeatures, RegressionLabels, LeastAngleRegression\n",
    "import numpy as np\n",
    "from time import time\n",
    "from sklearn.linear_model import LassoLars\n",
    "from sklearn.metrics import mean_squared_error as mse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "ran1=[30]\n",
    "ran2=[300]#,100000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "num_feat=10\n",
    "num_vec=100\n",
    "timesg=[]\n",
    "timesk=[]\n",
    "outsg=[]\n",
    "outsk=[]\n",
    "\n",
    "\n",
    "def gen_data(num_feat, num_vec):\n",
    "    X=np.random.rand(num_vec, num_feat)\n",
    "    xtest=np.random.rand(num_vec/10, num_feat)\n",
    "    \n",
    "    w=np.zeros((1, num_feat))\n",
    "\n",
    "    var=np.array(range(0, num_feat))\n",
    "    var=np.random.permutation(var)\n",
    "\n",
    "    for i in range(num_feat/2):\n",
    "        w[0][var[i]]=np.random.randint(100)\n",
    "\n",
    "    y=np.dot(w,X.T)\n",
    "    ytest=np.dot(w, xtest.T)\n",
    "    y=np.array(y)\n",
    "    y=np.reshape(y,(num_vec,))\n",
    "    ytest=np.reshape(ytest,(num_vec/10,))\n",
    "    return X, y, xtest, ytest\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "\n",
    "def run():\n",
    "    for f in ran1:\n",
    "        for v in ran2:\n",
    "            X,y, xtest, ytest =gen_data(f,v)\n",
    "            feat = RealFeatures(X.T)\n",
    "            lab = RegressionLabels(y)    \n",
    "            ftest = RealFeatures(xtest.T)\n",
    "            ltest = RegressionLabels(ytest) \n",
    "            \n",
    "            t1=time()\n",
    "            lambda1=0.01\n",
    "            modelsg = LeastAngleRegression()\n",
    "            modelsg.set_max_l1_norm(lambda1)\n",
    "            modelsg.set_labels(lab)\n",
    "            modelsg.train(feat)\n",
    "            t2=time()\n",
    "\n",
    "            timesg.append(t2-t1)\n",
    "            t3=time()\n",
    "            modelsk = LassoLars(alpha=lambda1)\n",
    "            modelsk.fit(X, y)\n",
    "            #out = model.coef\n",
    "            t4=time()\n",
    "            timesk.append(t4-t3)\n",
    "            sgout=modelsg.apply(ftest)\n",
    "            skout=modelsk.predict(xtest)\n",
    "            \n",
    "            sgout1=sgout.get_labels()\n",
    "            print sgout1, skout, ytest\n",
    "            outsg.append(mse(ytest, sgout1))\n",
    "            outsk.append(mse(ytest, skout))\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 385.72652834  385.72484367  385.72638108  385.72121201  385.71863408\n",
      "  385.72195451  385.71981821  385.72221152  385.72321434  385.72273265\n",
      "  385.72188016  385.71937371  385.7275607   385.71864802  385.71961235\n",
      "  385.72395046  385.71860946  385.72377739  385.71821185  385.71892004\n",
      "  385.72427833  385.72588812  385.72738081  385.72476188  385.72621216\n",
      "  385.72158667  385.72294017  385.72199666  385.7268188   385.72685648] [ 425.75474947  442.53869976  321.77677581  353.07133606  363.24071599\n",
      "  437.0867815   411.56535414  409.15885936  364.21720537  431.41260143\n",
      "  407.19071325  351.41136834  333.21237432  370.62953131  382.19169385\n",
      "  447.80262874  428.96436776  408.35301406  428.3309812   513.57059017\n",
      "  344.16988754  507.08600343  429.48442884  384.75214797  508.03107014\n",
      "  493.9912634   289.45732391  294.71884688  381.75522255  409.03580245] [ 426.13074984  442.87998496  321.17438232  352.66419854  363.22309136\n",
      "  437.54651105  412.32216736  409.4405144   363.91208429  431.76318783\n",
      "  407.45448377  351.28938453  332.62279038  370.52934846  381.84746235\n",
      "  448.20390181  428.99022733  408.80286699  429.48047462  514.83201116\n",
      "  343.59235217  508.02597633  429.99167192  384.5365857   509.06755368\n",
      "  494.97747608  288.04436562  294.06963151  381.54572513  408.85673744]\n"
     ]
    }
   ],
   "source": [
    "run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------\n",
      "\n",
      "N: 300 D: 30 shogun: 0.000226 , mse: 3705.699514 | scikit: 0.003624 , mse: 0.392170 \n"
     ]
    }
   ],
   "source": [
    "count=0\n",
    "for i in ran1:\n",
    "    for j in ran2:\n",
    "        print \"------------\\n\"\n",
    "        print 'N:', j , 'D:', i, 'shogun: %.6f , mse: %.6f | scikit: %.6f , mse: %.6f ' % (timesg[count], outsg[count]\\\n",
    "                                                                                           ,timesk[count], outsk[count])\n",
    "        count+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}