experiments.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import dio, util\n",
    "import numpy as np\n",
    "from sklearn.feature_extraction import DictVectorizer\n",
    "from collections import defaultdict"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Read in  data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reading data... \n",
      "Fitting input vectorizer with 'ccg' training data...\n",
      "\n",
      "Fitting input vectorizer with 'chunking' training data...\n",
      "\n",
      "Fitting input vectorizer with 'com' training data...\n",
      "\n",
      "Fitting input vectorizer with 'fnt' training data...\n",
      "\n",
      "Fitting input vectorizer with 'gpos' training data...\n",
      "\n",
      "Fitting input vectorizer with 'hyperlinks' training data...\n",
      "\n",
      "Fitting input vectorizer with 'keysem17' training data...\n",
      "\n",
      "Fitting input vectorizer with 'mwe' training data...\n",
      "\n",
      "Fitting input vectorizer with 'semcor' training data...\n",
      "\n",
      "Fitting input vectorizer with 'streusle' training data...\n",
      "Done!\n"
     ]
    }
   ],
   "source": [
    "tasks = util.read_tasks_config('tasks.cfg')\n",
    "# tasks = {'com': tasks['com'], 'fn': tasks['fn']}\n",
    "# dio.load_clusters('w2c.10240.txt', 'in')\n",
    "data = dio.read_data('data.cfg', input_type=dio.INDICES, tasks=tasks)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Vectorize all input data and labels\n",
    "\n",
    "Do this for all tasls. Building one matrix for some tasks means we have to pad the input data. Using default length 50 here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(39604, 31347)\n",
      "(15465, 28508)\n",
      "(3312, 5360)\n",
      "(3711, 8233)\n",
      "(3312, 5360)\n",
      "(2398, 6260)\n",
      "(2000, 9189)\n",
      "(9600, 24253)\n",
      "(1002, 4467)\n",
      "(8936, 14840)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "cv=CountVectorizer()\n",
    "\n",
    "# train data\n",
    "xs, ys, bows = {}, {}, {}\n",
    "for task, task_cfg in tasks.items():\n",
    "    x = data.input_vectorizer.transform(data.corpora[task]['train']['in'], pad_len=30)\n",
    "    y = data.output_vectorizers[task].transform(data.corpora[task]['train']['out'], pad_len=30)\n",
    "    xs[task] = x\n",
    "    ys[task] = y\n",
    "    bows[task] = cv.fit_transform([' '.join(sent) for sent in data.corpora[task]['train']['in']])\n",
    "    \n",
    "# dev data\n",
    "xs_dev, ys_dev = {}, {}\n",
    "for task, task_cfg in tasks.items():\n",
    "    x = data.input_vectorizer.transform(data.corpora[task]['dev']['in'], pad_len=30)\n",
    "    x = data.output_vectorizers[task].transform(data.corpora[task]['dev']['out'], pad_len=30)\n",
    "    xs_dev[task] = x\n",
    "    ys_dev[task] = y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute word distributions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "from scipy.stats import entropy\n",
    "\n",
    "# http://stackoverflow.com/questions/15880133/jensen-shannon-divergence\n",
    "def JSD(P, Q):\n",
    "    _P = P / np.linalg.norm(P, ord=1)\n",
    "    _Q = Q / np.linalg.norm(Q, ord=1)\n",
    "    _M = 0.5 * (_P + _Q)\n",
    "    return 0.5 * (entropy(_P, _M) + entropy(_Q, _M))\n",
    "    \n",
    "    \n",
    "voc_len = 10242\n",
    "jsds = {}\n",
    "\n",
    "for task in tasks:\n",
    "    c = Counter(xs[task].flat)\n",
    "    dist_tr = np.array([c[i] for i in range(voc_len)])\n",
    "    c_dev = Counter(xs_dev[task].flat)\n",
    "    dist_dev = np.array([c_dev[i] for i in range(voc_len)])    \n",
    "    jsds[task] = JSD(dist_tr, dist_dev)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### Single-task model parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 318,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFkCAYAAAAJ/rPWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wCy1tjZDbfnzwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f460efe6470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from scipy.optimize import curve_fit\n",
    "\n",
    "def read_learning_curve(f):\n",
    "    iter_loss = []\n",
    "    for line in open(f):\n",
    "        spl = line.split()\n",
    "        iter_loss.append((int(spl[1].strip(',')), float(spl[5].strip(','))))\n",
    "    return iter_loss\n",
    "\n",
    "def func(x, a, b, c, d):\n",
    "    return a + b * np.log(np.multiply(x,c)+d)\n",
    "\n",
    "def get_curve_params(task):\n",
    "    xs, ys = zip(*read_learning_curve('logs/{}/loss_tr.log'.format(task)))\n",
    "    popt, _ = curve_fit(func, xs[:5000:10], ys[:5000:10])\n",
    "    return popt[1], popt[2]\n",
    "\n",
    "def get_curve_gradients(task, pos=[0.001, 0.4, 0.6, 0.8]):\n",
    "    def get_gradient_at(xs, ys, i):\n",
    "        xbefore = xs[i-1]\n",
    "        ybefore = ys[i-1]\n",
    "        xafter = xs[i+1]\n",
    "        yafter = ys[i+1]\n",
    "        return (yafter-ybefore)/(xafter-xbefore)\n",
    "    \n",
    "    xs, ys = zip(*read_learning_curve('logs/{}/loss_tr.log'.format(task)))\n",
    "    gradients = []\n",
    "    for p in pos:\n",
    "        gradients.append(get_gradient_at(xs, ys, int(len(xs)*p)))\n",
    "    return gradients\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "iters, mwe_loss = zip(*read_learning_curve(\"logs/mwe/loss_tr.log\"))\n",
    "iters, fnt_loss = zip(*read_learning_curve(\"logs/fnt/loss_tr.log\"))\n",
    "iters, key_loss = zip(*read_learning_curve(\"logs/keysem17/loss_tr.log\"))\n",
    "iters, gpos_loss = zip(*read_learning_curve(\"logs/gpos/loss_tr.log\"))\n",
    "iters, hyp_loss = zip(*read_learning_curve(\"logs/hyperlinks/loss_tr.log\"))\n",
    "iters, ccg_loss = zip(*read_learning_curve(\"logs/ccg/loss_tr.log\"))\n",
    "iters, com_loss = zip(*read_learning_curve(\"logs/com/loss_tr.log\"))\n",
    "iters, str_loss = zip(*read_learning_curve(\"logs/streusle/loss_tr.log\"))\n",
    "iters, sem_loss = zip(*read_learning_curve(\"logs/semcor/loss_tr.log\"))\n",
    "iters, chu_loss = zip(*read_learning_curve(\"logs/chunking/loss_tr.log\"))\n",
    "\n",
    "plt.plot(mwe_loss[:5000:10], label='MWE')\n",
    "plt.plot(fnt_loss[:5000:10], label='FNT')\n",
    "plt.plot(key_loss[:5000:10], label='KEY')\n",
    "plt.plot(gpos_loss[:5000:10], label=\"POS\")\n",
    "plt.plot(hyp_loss[:5000:10], label=\"HYP\")\n",
    "plt.plot(ccg_loss[:5000:10], label=\"CCG\")\n",
    "plt.plot(chu_loss[:5000:10], label=\"CHU\")\n",
    "plt.plot(com_loss[:5000:10], label=\"COM\")\n",
    "plt.plot(str_loss[:5000:10], label=\"STR\")\n",
    "plt.plot(sem_loss[:5000:10], label=\"SEM\")\n",
    "\n",
    "plt.legend()\n",
    "plt.savefig('loss.png')\n",
    "\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Collect features for every task/dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "400004\n"
     ]
    }
   ],
   "source": [
    "voc = set([l.strip() for l in open(\"emb_voc.txt\")])\n",
    "print(len(voc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=== ccg ===\n",
      "-0.006298333406448364\n",
      "\n",
      "=== chunking ===\n",
      "0.001435372233390808\n",
      "\n",
      "=== com ===\n",
      "5.260854959487915e-05\n",
      "\n",
      "=== fnt ===\n",
      "4.1321665048599246e-05\n",
      "\n",
      "=== gpos ===\n",
      "-0.0006624087691307068\n",
      "\n",
      "=== hyperlinks ===\n",
      "-0.0008521348237991333\n",
      "\n",
      "=== keysem17 ===\n",
      "-9.096264839172363e-05\n",
      "\n",
      "=== mwe ===\n",
      "-7.283687591552734e-05\n",
      "\n",
      "=== semcor ===\n",
      "0.0011842578649520874\n",
      "\n",
      "=== streusle ===\n",
      "0.00032129883766174316\n",
      "\n"
     ]
    }
   ],
   "source": [
    "featuredicts = {}\n",
    "import scipy\n",
    "\n",
    "def get_types_tokens(x):\n",
    "    tokens = [w for w in x.flat if w > 1]  # indices 0 and 1 are reserved for PAD and GO\n",
    "    return set(tokens), tokens\n",
    "\n",
    "for task in sorted(tasks):\n",
    "    labels = [y for y in ys[task].flat if y>3]\n",
    "    D = {}\n",
    "    print(\"=== {} ===\".format(task))\n",
    "    # nuclear norm: sum of singular values\n",
    "    D[\"fro_norm\"] = scipy.sparse.linalg.norm(bows[task], ord='fro')\n",
    "    # length of dataset\n",
    "    D[\"length\"] = xs[task].shape[0]\n",
    "    # number of labels\n",
    "    D[\"n_labels\"] = tasks[task].get_num_labels()-4\n",
    "    # labels entropy\n",
    "    D[\"label_entropy\"] = entropy([count for lbl, count in Counter(labels).items()])\n",
    "    # token/type ratio\n",
    "    types, tokens = get_types_tokens(xs[task])\n",
    "    D[\"tokens/types\"] = len(tokens)/len(types)\n",
    "    # BOW JS divergence train/test\n",
    "    D[\"js_div\"] = jsds[task]\n",
    "    # OOV rate\n",
    "    all_words = [w for sent in data.corpora[task]['train']['in'] for w in sent]\n",
    "    D[\"oov_rate\"] = len([w for w in all_words if w not in voc]) / len(all_words) * 100\n",
    "    # Learning curve gradients\n",
    "    grads = get_curve_gradients(task, [0.1, 0.2, 0.3, 0.5, 0.7])\n",
    "    D[\"loss_grad_10\"] = grads[0]\n",
    "    D[\"loss_grad_20\"] = grads[1]\n",
    "    D[\"loss_grad_30\"] = grads[2]\n",
    "    D[\"loss_grad_50\"] = grads[3]\n",
    "    D[\"loss_grad_70\"] = grads[4]\n",
    "    params = get_curve_params(task)\n",
    "    D[\"curve_steepness0\"] = params[0]\n",
    "    D[\"curve_steepness1\"] = params[1]\n",
    "    # add to featuredicts\n",
    "    print(D[\"loss_grad_30\"])\n",
    "    print()\n",
    "    featuredicts[task] = D "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pair tasks\n",
    "Compute features for pairs of tasks, where one task is treated as main and the other as auxiliary. A single instance in this set is a concatenation of the features collected above for the main task, the aux task, and the ratio of the two."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "90\n"
     ]
    }
   ],
   "source": [
    "v = DictVectorizer()\n",
    "paired_feats = {}\n",
    "\n",
    "for main_task, main_task_feats in featuredicts.items():\n",
    "    for aux_task, aux_task_feats in featuredicts.items():\n",
    "        if main_task == aux_task: continue\n",
    "        D = {}\n",
    "        for feat in main_task_feats.keys():\n",
    "            D[\"main_\"+feat] = main_task_feats[feat]\n",
    "            D[\"aux_\"+feat] = aux_task_feats[feat]\n",
    "            D[\"main/aux_\"+feat] = main_task_feats[feat] / aux_task_feats[feat]\n",
    "        paired_feats[main_task+\"_\"+aux_task] = D\n",
    "\n",
    "print(len(paired_feats))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute performance differences between STL/MTL\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GAINS (40)\n",
      "[('hyperlinks_semcor', 23.200162754621978), ('hyperlinks_gpos', 22.243979350507324), ('mwe_chunking', 19.564614531336645), ('mwe_ccg', 17.979293769643178), ('mwe_semcor', 15.977999630245886), ('hyperlinks_ccg', 12.0514711492002), ('hyperlinks_fnt', 9.175291813951114), ('mwe_streusle', 8.610648918469233), ('hyperlinks_streusle', 8.086870279480207), ('mwe_com', 7.378905527824009), ('keysem17_ccg', 5.653595277566903), ('mwe_fnt', 5.502403401737843), ('gpos_ccg', 4.8887059877267935), ('hyperlinks_chunking', 4.018004730056202), ('gpos_semcor', 3.363633062612714), ('keysem17_chunking', 3.21491098957285), ('gpos_chunking', 2.8746414263063125), ('gpos_streusle', 2.8673791742819477), ('gpos_com', 1.8712402716082321), ('ccg_gpos', 1.8078938501399966), ('streusle_semcor', 1.7171010910746576), ('mwe_gpos', 1.63847291551118), ('streusle_chunking', 1.4585603017935211), ('hyperlinks_keysem17', 1.454619433919091), ('ccg_chunking', 1.3803199832688406), ('ccg_semcor', 1.3652154716674136), ('com_chunking', 1.338994121489212), ('gpos_mwe', 1.2890497343226226), ('com_fnt', 1.2777596342259834), ('gpos_fnt', 0.898098500344946), ('ccg_streusle', 0.8365575656174906), ('com_mwe', 0.8164598301763526), ('keysem17_streusle', 0.6894628387276835), ('keysem17_semcor', 0.5887547836326279), ('ccg_fnt', 0.5774570973776205), ('ccg_com', 0.4542972335505935), ('ccg_mwe', 0.4508115770271992), ('ccg_keysem17', 0.3044140030441289), ('ccg_hyperlinks', 0.23818652909941562), ('streusle_gpos', 0.03739898209726977)]\n",
      "\n",
      "\n",
      "LOSSES (50)\n",
      "[('chunking_ccg', -0.05218539660901689), ('chunking_semcor', -0.09904575274771732), ('chunking_fnt', -0.11608588225271443), ('chunking_com', -0.1512311493567342), ('chunking_keysem17', -0.2247167078469836), ('streusle_com', -0.2601668319810124), ('gpos_keysem17', -0.26144107287668916), ('chunking_mwe', -0.26731703160943754), ('chunking_streusle', -0.32163244440658323), ('keysem17_fnt', -0.43226995956183734), ('chunking_gpos', -0.45475845616426147), ('chunking_hyperlinks', -0.5005538042089142), ('com_streusle', -0.6327563683866888), ('fnt_streusle', -0.6755243068121031), ('streusle_fnt', -0.715458787947787), ('semcor_chunking', -0.7614284907046409), ('semcor_fnt', -0.8136972368197992), ('semcor_keysem17', -0.8348872690286502), ('gpos_hyperlinks', -0.8496834868492731), ('semcor_gpos', -0.854664632423574), ('fnt_chunking', -1.0115631839252681), ('keysem17_com', -1.03341958074461), ('semcor_mwe', -1.0849296490930582), ('semcor_com', -1.236085212182847), ('keysem17_gpos', -1.2503292378724162), ('semcor_hyperlinks', -1.2714019325309356), ('com_gpos', -1.3532821685173113), ('streusle_keysem17', -1.4016488072976951), ('streusle_hyperlinks', -1.4715686433925734), ('streusle_mwe', -1.635798956080603), ('semcor_streusle', -1.707916596033221), ('streusle_ccg', -1.7268573472739468), ('com_hyperlinks', -2.3861038536904022), ('keysem17_hyperlinks', -2.638551043490389), ('fnt_semcor', -2.9588653948887655), ('com_semcor', -2.9882429784454656), ('fnt_keysem17', -3.560288821106694), ('mwe_hyperlinks', -3.8084673691994864), ('keysem17_mwe', -4.731729234773708), ('com_keysem17', -4.841606792945797), ('semcor_ccg', -4.993784257218737), ('com_ccg', -5.031433703461785), ('fnt_hyperlinks', -5.724723844974067), ('fnt_ccg', -5.78503851522515), ('mwe_keysem17', -5.849047883157699), ('fnt_com', -6.110737734580973), ('hyperlinks_mwe', -7.7130432571268726), ('fnt_mwe', -9.362560099260708), ('fnt_gpos', -9.410811835461574), ('hyperlinks_com', -10.696030313048354)]\n",
      "\n",
      "\n",
      "Most helpful tasks\n",
      "Counter({'chunking': 7, 'semcor': 6, 'fnt': 5, 'streusle': 5, 'ccg': 4, 'gpos': 4, 'mwe': 3, 'com': 3, 'keysem17': 2, 'hyperlinks': 1})\n"
     ]
    }
   ],
   "source": [
    "def get_f1(file, task, iters):\n",
    "    \"\"\"\n",
    "    gets accuracy of specfic task in loss_dv logfile at iters iterations\n",
    "    \"\"\"\n",
    "    f = open(file)\n",
    "    for line in f:\n",
    "        line = line.strip().replace(',', '').split()\n",
    "        if int(line[1]) == iters:\n",
    "            if line[3] == task:\n",
    "                acc = float(line[9])\n",
    "                break\n",
    "    f.close()\n",
    "    return acc\n",
    "\n",
    "# collect all accuracies after 50k batches for single tasks and 100k batches for MTL exps\n",
    "accs = {}\n",
    "diffs = {}\n",
    "for task1 in tasks.keys():\n",
    "    accs[task1] = get_f1('logs/{}/loss_dv.log'.format(task1), task1 , 50000)\n",
    "    for task2 in tasks.keys():\n",
    "        if task1 == task2: continue\n",
    "        exppair = '_'.join(sorted([task1, task2]))\n",
    "        accs[task1+'_'+task2] = get_performance('logs/{}/loss_dv.log'.format(exppair), task1 , 100000)\n",
    "        # accuracy ratio between STL and MTL\n",
    "        diffs[task1+'_'+task2] = (accs[task1+'_'+task2] / accs[task1]  -1) * 100\n",
    "        \n",
    "# print(diffs)\n",
    "gains = sorted([(pair, diff) for pair, diff in diffs.items() if diff > 0], key=lambda x: x[1], reverse=True)\n",
    "losses = sorted([(pair, diff) for pair, diff in diffs.items() if diff <= 0], key=lambda x: x[1], reverse=True)\n",
    "\n",
    "print(\"GAINS ({})\".format(len(gains)))\n",
    "print(gains)\n",
    "print(\"\\n\\nLOSSES ({})\".format(len(losses)))\n",
    "print(losses)\n",
    "\n",
    "print(\"\\n\\nMost helpful tasks\")\n",
    "aux_task_helpfulness = Counter([task.split('_')[1] for task, diff in gains])\n",
    "print(aux_task_helpfulness)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Analyzing task contribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 306,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['ccg', 'chunking', 'com', 'fnt', 'gpos', 'hyperlinks', 'keysem17', 'mwe', 'semcor', 'streusle']\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1cAAALGCAYAAACpj2uiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAQ2uRERERERELKDBlYiIiIiIiAX+C69uKZ6gOO43AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f45ec13a080>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "matrix = np.zeros([len(tasks), len(tasks)])\n",
    "for i, task1 in enumerate(sorted(tasks)):\n",
    "    for j, task2 in enumerate(sorted(tasks)):\n",
    "        if not task1 == task2:\n",
    "            matrix[i,j] = diffs[task1+'_'+task2]\n",
    "        \n",
    "fig, ax = plt.subplots(figsize=(15,12))  \n",
    "sns.set(font_scale=2)\n",
    "\n",
    "# mask diagonal\n",
    "mask = np.zeros_like(matrix)\n",
    "for diag in range(len(mask)):\n",
    "    mask[diag,diag] = True\n",
    "\n",
    "heatmap = sns.heatmap(matrix, mask=mask, annot=True, ax=ax, cmap=\"RdBu_r\")\n",
    "ax.xaxis.set_ticks_position('top')\n",
    "print(sorted(tasks))\n",
    "labels=\"CCG Chunking Compression FNT POS Hyperlinks Keyphrases MWE Semcor Streusle\".split()\n",
    "xlabels=\"CCG CHU COM FNT POS HYP KEY MWE Sem Str\".upper().split()\n",
    "ax.set_xticklabels(xlabels, minor=False, rotation=0)\n",
    "ax.set_yticklabels(xlabels[::-1], minor=False, rotation=0)\n",
    "plt.show()\n",
    "fig.savefig('heatmap.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prediction\n",
    "Try to predict helpfulness of MTL based on differences in performance, using logistic regression\n",
    "\n",
    "### Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction import DictVectorizer\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "v = DictVectorizer(sparse=False)\n",
    "\n",
    "X, y = [], []\n",
    "xs_dicts=[]\n",
    "pair_seq = [p for p in paired_feats.keys()]\n",
    "for pair in pair_seq:\n",
    "    xs_dicts.append(paired_feats[pair])\n",
    "    y.append(diffs[pair])\n",
    "X = v.fit_transform(xs_dicts)\n",
    "\n",
    "X = MinMaxScaler().fit_transform(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classification\n",
    "### All features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.888888888889\n",
      "[(36, 'main_loss_grad_30', 1.0), (4, 'aux_label_entropy', 3.3999999999999999), (29, 'main_curve_steepness1', 3.71), (35, 'main_loss_grad_20', 4.7999999999999998), (40, 'main_oov_rate', 6.1900000000000004), (8, 'aux_loss_grad_30', 6.29), (37, 'main_loss_grad_50', 7.4000000000000004), (26, 'main/aux_oov_rate', 7.5300000000000002), (39, 'main_n_labels', 8.0899999999999999), (30, 'main_fro_norm', 10.970000000000001), (18, 'main/aux_label_entropy', 12.539999999999999), (7, 'aux_loss_grad_20', 14.32), (25, 'main/aux_n_labels', 14.460000000000001), (38, 'main_loss_grad_70', 14.529999999999999), (2, 'aux_fro_norm', 15.69), (0, 'aux_curve_steepness0', 18.870000000000001), (41, 'main_tokens/types', 19.030000000000001), (10, 'aux_loss_grad_70', 20.02), (13, 'aux_tokens/types', 20.699999999999999), (6, 'aux_loss_grad_10', 20.829999999999998), (16, 'main/aux_fro_norm', 22.300000000000001), (14, 'main/aux_curve_steepness0', 22.620000000000001), (5, 'aux_length', 23.760000000000002), (15, 'main/aux_curve_steepness1', 25.690000000000001), (1, 'aux_curve_steepness1', 25.949999999999999), (11, 'aux_n_labels', 26.789999999999999), (3, 'aux_js_div', 26.850000000000001), (12, 'aux_oov_rate', 30.359999999999999), (32, 'main_label_entropy', 30.420000000000002), (31, 'main_js_div', 31.039999999999999), (27, 'main/aux_tokens/types', 31.329999999999998), (9, 'aux_loss_grad_50', 31.77), (34, 'main_loss_grad_10', 31.93), (21, 'main/aux_loss_grad_20', 32.57), (23, 'main/aux_loss_grad_50', 32.770000000000003), (28, 'main_curve_steepness0', 33.009999999999998), (17, 'main/aux_js_div', 34.619999999999997), (20, 'main/aux_loss_grad_10', 34.649999999999999), (33, 'main_length', 34.710000000000001), (24, 'main/aux_loss_grad_70', 35.049999999999997), (22, 'main/aux_loss_grad_30', 35.869999999999997), (19, 'main/aux_length', 38.57)]\n",
      "\n",
      "main_loss_grad_30 -1.56591577373\n",
      "main_loss_grad_20 -1.16401902759\n",
      "main_curve_steepness1 1.00650006066\n",
      "main_n_labels 0.827748270865\n",
      "aux_label_entropy 0.797912243925\n",
      "aux_loss_grad_30 0.790909442405\n",
      "main_loss_grad_50 0.781292181767\n",
      "main_oov_rate 0.697454036103\n",
      "main/aux_oov_rate 0.678210091847\n",
      "aux_loss_grad_20 0.575789173046\n",
      "main_fro_norm -0.516229201808\n",
      "main/aux_n_labels 0.503641505604\n",
      "main_loss_grad_70 0.433715622821\n",
      "main/aux_label_entropy -0.411204045104\n",
      "aux_fro_norm 0.345581015129\n",
      "main_tokens/types -0.297074349252\n",
      "aux_curve_steepness0 -0.296744797375\n",
      "aux_loss_grad_70 -0.278951968956\n",
      "aux_loss_grad_10 0.267856039213\n",
      "aux_tokens/types 0.254494707253\n",
      "main/aux_curve_steepness0 -0.241463228192\n",
      "aux_length 0.237413962388\n",
      "main/aux_fro_norm -0.233753741254\n",
      "aux_js_div -0.207247738018\n",
      "aux_n_labels -0.183862949662\n",
      "aux_curve_steepness1 -0.174437854792\n",
      "main/aux_tokens/types -0.117499121484\n",
      "main/aux_curve_steepness1 -0.103749705592\n",
      "main/aux_loss_grad_20 0.10355837316\n",
      "main_label_entropy -0.102060153211\n",
      "aux_loss_grad_50 -0.098584543925\n",
      "main/aux_loss_grad_50 0.0768257161858\n",
      "aux_oov_rate 0.0613135144519\n",
      "main/aux_loss_grad_30 -0.0600558785188\n",
      "main_length -0.0325710178819\n",
      "main_curve_steepness0 0.0274721114683\n",
      "main/aux_loss_grad_10 0.0233539303581\n",
      "main_js_div 0.0193026190686\n",
      "main/aux_js_div -0.014779005837\n",
      "main_loss_grad_10 0.00689890898435\n",
      "main/aux_length -0.00655948336576\n",
      "main/aux_loss_grad_70 -0.000446438752313\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression, LogisticRegression\n",
    "import numpy as np\n",
    "y_log = np.sign(y)\n",
    "lr = LogisticRegression()\n",
    "lr.fit(X, y_log)\n",
    "print(lr.score(X, y_log))\n",
    "# print(lr.score(X_train, y_train))\n",
    "\n",
    "from sklearn.model_selection import cross_val_score, KFold, train_test_split\n",
    "from sklearn.feature_selection import RFE\n",
    "\n",
    "feature_ranks = defaultdict(list)\n",
    "feature_coeffs = defaultdict(list)\n",
    "for _ in range(100):\n",
    "    X_tr, X_te, y_tr, y_te = train_test_split(X, y_log, test_size=0.2, random_state=np.random.randint(100))\n",
    "    rfe = RFE(lr, len(lr.coef_))\n",
    "    rfe.fit(X_tr, y_tr)\n",
    "    lr = LogisticRegression()\n",
    "    lr.fit(X_tr, y_tr)\n",
    "    for feat_id, rank in enumerate(rfe.ranking_):\n",
    "        feature_ranks[feat_id].append(rank)\n",
    "        feature_coeffs[feat_id].append(lr.coef_[0][feat_id])\n",
    "\n",
    "ranking = sorted([(feat_id, v.feature_names_[feat_id], np.array(ranks).mean()) \n",
    "                  for feat_id, ranks in feature_ranks.items()], key=lambda x: x[2])\n",
    "print(ranking)\n",
    "\n",
    "coefranking = sorted([(feat_id, v.feature_names_[feat_id], np.array(coef).mean()) \n",
    "                  for feat_id, coef in feature_coeffs.items()], key=lambda x: abs(x[2]), reverse=True)\n",
    "\n",
    "print()\n",
    "for fid, fname, coef in coefranking:\n",
    "    print(fname, coef)\n",
    "\n",
    "# print(rfe.score(X, y_log))\n",
    "\n",
    "# TODO fit RFE in several iterations with random train/test splits, record avg rank of each feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "All features\n",
      "0.668625222645 0.0412209332315\n",
      "\n",
      "19 main/aux_length\n",
      "0.672778584367 0.0409667625765\n",
      "\n",
      "22 main/aux_loss_grad_30\n",
      "0.673839034976 0.0435939859929\n",
      "\n",
      "24 main/aux_loss_grad_70\n",
      "0.671125265687 0.0437586251408\n",
      "\n",
      "33 main_length\n",
      "0.671656514435 0.0392643689153\n",
      "\n",
      "20 main/aux_loss_grad_10\n",
      "0.676262389245 0.0404401065145\n",
      "\n",
      "17 main/aux_js_div\n",
      "0.670101483656 0.044942303407\n",
      "\n",
      "28 main_curve_steepness0\n",
      "0.671069056692 0.0431601634108\n",
      "\n",
      "23 main/aux_loss_grad_50\n",
      "0.674050235334 0.0415769385401\n",
      "\n",
      "21 main/aux_loss_grad_20\n",
      "0.675816822445 0.0396203149333\n",
      "\n",
      "34 main_loss_grad_10\n",
      "0.670076958284 0.0374741688659\n",
      "\n",
      "9 aux_loss_grad_50\n",
      "0.679906826387 0.0429277208858\n",
      "\n",
      "27 main/aux_tokens/types\n",
      "0.679281153539 0.0446389381068\n",
      "\n",
      "31 main_js_div\n",
      "0.678034234221 0.0361681404076\n",
      "\n",
      "32 main_label_entropy\n",
      "0.688768329658 0.0416889292191\n",
      "\n",
      "12 aux_oov_rate\n",
      "0.694478887298 0.0407112559679\n",
      "\n",
      "3 aux_js_div\n",
      "0.693856871989 0.0382382626645\n",
      "\n",
      "11 aux_n_labels\n",
      "0.699347698603 0.0371473571082\n",
      "\n",
      "1 aux_curve_steepness1\n",
      "0.699063154407 0.0443672399551\n",
      "\n",
      "15 main/aux_curve_steepness1\n",
      "0.698414107375 0.043299989879\n",
      "\n",
      "5 aux_length\n",
      "0.70098007115 0.0425513757249\n",
      "\n",
      "14 main/aux_curve_steepness0\n",
      "0.699936718596 0.0408650279737\n",
      "\n",
      "16 main/aux_fro_norm\n",
      "0.698075908526 0.041156084167\n",
      "\n",
      "6 aux_loss_grad_10\n",
      "0.712888710137 0.039898063774\n",
      "\n",
      "13 aux_tokens/types\n",
      "0.698163378177 0.0394731552452\n",
      "\n",
      "10 aux_loss_grad_70\n",
      "0.711179357261 0.0426003924344\n",
      "\n",
      "41 main_tokens/types\n",
      "0.703611714429 0.0421654857685\n",
      "\n",
      "0 aux_curve_steepness0\n",
      "0.701118328163 0.0403420366841\n",
      "\n",
      "2 aux_fro_norm\n",
      "0.703759731823 0.0425581910619\n",
      "\n",
      "38 main_loss_grad_70\n",
      "0.682757366936 0.0374121490506\n",
      "\n",
      "25 main/aux_n_labels\n",
      "0.687128566738 0.0458645120702\n",
      "\n",
      "7 aux_loss_grad_20\n",
      "0.683406691898 0.039013889967\n",
      "\n",
      "18 main/aux_label_entropy\n",
      "0.685253381328 0.0408153431402\n",
      "\n",
      "30 main_fro_norm\n",
      "0.651262594045 0.0399942914399\n",
      "\n",
      "39 main_n_labels\n",
      "0.659093773409 0.0391950250999\n",
      "\n",
      "26 main/aux_oov_rate\n",
      "0.635731621337 0.0483719723593\n",
      "\n",
      "37 main_loss_grad_50\n",
      "0.636600114866 0.0482739299056\n",
      "\n",
      "8 aux_loss_grad_30\n",
      "0.609603344179 0.0294886479974\n",
      "\n",
      "40 main_oov_rate\n",
      "0.548752182458 0.0340783537113\n",
      "\n",
      "35 main_loss_grad_20\n",
      "0.574143225024 0.0405724362146\n",
      "\n",
      "29 main_curve_steepness1\n",
      "0.533835892625 0.0405329150791\n",
      "\n",
      "4 aux_label_entropy\n",
      "0.354538417138 0.0237657759299\n",
      "\n",
      "36 main_loss_grad_30\n",
      "0.00285714285714 0.0124539969815\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "X_abl = X\n",
    "means = []\n",
    "print(\"All features\")\n",
    "for _ in range(100):\n",
    "    lr = LogisticRegression()\n",
    "    shuffle = KFold(n_splits=5, shuffle=True, random_state=np.random.randint(100))\n",
    "    scores = cross_val_score(lr, X_abl, y_log, cv=shuffle, scoring='f1')\n",
    "    means.append(scores.mean())\n",
    "means= np.array(means)\n",
    "print(means.mean(), means.std())\n",
    "print()\n",
    "\n",
    "for feat_id, feat_name, avg_rank in ranking[::-1]:\n",
    "    for x in X_abl:\n",
    "        x[feat_id] = 0\n",
    "    print(feat_id, feat_name)\n",
    "    means = []\n",
    "    for _ in range(100):\n",
    "        lr = LogisticRegression()\n",
    "        shuffle = KFold(n_splits=5, shuffle=True, random_state=np.random.randint(100))\n",
    "        scores = cross_val_score(lr, X_abl, y_log, cv=shuffle, scoring='f1')\n",
    "        means.append(scores.mean())\n",
    "    means= np.array(means)\n",
    "    print(means.mean(), means.std())\n",
    "    print()\n",
    "\n",
    "# feats = sorted([(f, lr.coef_[0][i]) for i, f in enumerate(v.feature_names_)], key=lambda x: abs(x[1]), reverse=True)\n",
    "# for f in feats:    \n",
    "#     print(f)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data features only"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction import DictVectorizer\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "v = DictVectorizer(sparse=False)\n",
    "\n",
    "X, y = [], []\n",
    "xs_dicts=[]\n",
    "pair_seq = [p for p in paired_feats.keys()]\n",
    "for pair in pair_seq:\n",
    "    x_dict = paired_feats[pair]\n",
    "    delete_keys= [k for k in x_dict.keys() if \"grad\" in k or \"curve\" in k]\n",
    "    for feat in delete_keys:\n",
    "            x_dict.pop(feat)\n",
    "    xs_dicts.append(x_dict)\n",
    "    y.append(diffs[pair])\n",
    "X = v.fit_transform(xs_dicts)\n",
    "\n",
    "X = MinMaxScaler().fit_transform(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.19918826  0.71029611  0.04986286  0.03616393  0.          0.27880768\n",
      "  0.29505596  0.2424462   0.23377364  0.5035391   0.16262989  0.05680705\n",
      "  0.05144628  0.15587143  0.5757352   0.57170327  0.61142554  0.37466971\n",
      "  0.05533905  0.29841724  0.39880945] (90, 21)\n",
      "[(0, 'aux_fro_norm', 1.0), (1, 'aux_js_div', 1.0), (2, 'aux_label_entropy', 1.0), (3, 'aux_length', 1.0), (4, 'aux_n_labels', 1.0), (5, 'aux_oov_rate', 1.0), (6, 'aux_tokens/types', 1.0), (7, 'main/aux_fro_norm', 1.0), (8, 'main/aux_js_div', 1.0), (9, 'main/aux_label_entropy', 1.0), (10, 'main/aux_length', 1.0), (11, 'main/aux_n_labels', 1.0), (12, 'main/aux_oov_rate', 1.0), (13, 'main/aux_tokens/types', 1.0), (14, 'main_fro_norm', 1.0), (15, 'main_js_div', 1.0), (16, 'main_label_entropy', 1.0), (17, 'main_length', 1.0), (18, 'main_n_labels', 1.0), (19, 'main_oov_rate', 1.0), (20, 'main_tokens/types', 1.0)]\n",
      "All features\n",
      "0.527321843791 0.0446344238586\n",
      "\n",
      "20 main_tokens/types\n",
      "0.540041421478 0.0479753267555\n",
      "\n",
      "19 main_oov_rate\n",
      "0.541509586922 0.0477780174441\n",
      "\n",
      "18 main_n_labels\n",
      "0.486114905579 0.0559587356725\n",
      "\n",
      "17 main_length\n",
      "0.441388662352 0.0742474051688\n",
      "\n",
      "16 main_label_entropy\n",
      "0.42828348741 0.0629873240803\n",
      "\n",
      "15 main_js_div\n",
      "0.403096075785 0.0700270753959\n",
      "\n",
      "14 main_fro_norm\n",
      "0.40731879548 0.0654882985652\n",
      "\n",
      "13 main/aux_tokens/types\n",
      "0.398933660577 0.0526633120439\n",
      "\n",
      "12 main/aux_oov_rate\n",
      "0.397822486921 0.066126106493\n",
      "\n",
      "11 main/aux_n_labels\n",
      "0.312498133892 0.0646213328638\n",
      "\n",
      "10 main/aux_length\n",
      "0.325108664847 0.0576736986887\n",
      "\n",
      "9 main/aux_label_entropy\n",
      "0.326156658598 0.0623444682391\n",
      "\n",
      "8 main/aux_js_div\n",
      "0.290640117416 0.0689038880472\n",
      "\n",
      "7 main/aux_fro_norm\n",
      "0.29580223069 0.0699162987181\n",
      "\n",
      "6 aux_tokens/types\n",
      "0.30696963903 0.067979702719\n",
      "\n",
      "5 aux_oov_rate\n",
      "0.311756378348 0.0557957340044\n",
      "\n",
      "4 aux_n_labels\n",
      "0.282102935609 0.0647695231216\n",
      "\n",
      "3 aux_length\n",
      "0.284849610417 0.0596069057979\n",
      "\n",
      "2 aux_label_entropy\n",
      "0.0942944252514 0.0516738269833\n",
      "\n",
      "1 aux_js_div\n",
      "0.0873567543568 0.044825485495\n",
      "\n",
      "0 aux_fro_norm\n",
      "0.00514285714286 0.0163532434529\n",
      "\n"
     ]
    }
   ],
   "source": [
    "X_abl = X\n",
    "means = []\n",
    "\n",
    "print(X_abl[0], X_abl.shape)\n",
    "\n",
    "from sklearn.model_selection import cross_val_score, KFold, train_test_split\n",
    "from sklearn.feature_selection import RFE\n",
    "\n",
    "feature_ranks = defaultdict(list)\n",
    "for _ in range(100):\n",
    "    lr = LogisticRegression()\n",
    "    X_tr, X_te, y_tr, y_te = train_test_split(X, y_log, test_size=0.2)\n",
    "    rfe = RFE(lr, 42)\n",
    "    rfe.fit(X_tr, y_tr)\n",
    "    for feat_id, rank in enumerate(rfe.ranking_):\n",
    "        feature_ranks[feat_id].append(rank)\n",
    "\n",
    "ranking = sorted([(feat_id, v.feature_names_[feat_id], np.array(ranks).mean()) \n",
    "                  for feat_id, ranks in feature_ranks.items()], key=lambda x: x[2])\n",
    "print(ranking)\n",
    "# print(rfe.score(X, y_log))\n",
    "            \n",
    "print(\"All features\")\n",
    "for _ in range(100):\n",
    "    lr = LogisticRegression()\n",
    "    shuffle = KFold(n_splits=5, shuffle=True, random_state=np.random.randint(100))\n",
    "    scores = cross_val_score(lr, X_abl, y_log, cv=shuffle, scoring='f1')\n",
    "    means.append(scores.mean())\n",
    "means= np.array(means)\n",
    "print(means.mean(), means.std())\n",
    "print()\n",
    "\n",
    "for feat_id, feat_name, avg_rank in ranking[::-1]:\n",
    "    for x in X_abl:\n",
    "        x[feat_id] = 0\n",
    "    print(feat_id, feat_name)\n",
    "    means = []\n",
    "    for _ in range(100):\n",
    "        lr = LogisticRegression()\n",
    "        shuffle = KFold(n_splits=5, shuffle=True, random_state=np.random.randint(100))\n",
    "        scores = cross_val_score(lr, X_abl, y_log, cv=shuffle, scoring='f1')\n",
    "        means.append(scores.mean())\n",
    "    means= np.array(means)\n",
    "    print(means.mean(), means.std())\n",
    "    print()\n",
    "\n",
    "# feats = sorted([(f, lr.coef_[0][i]) for i, f in enumerate(v.feature_names_)], key=lambda x: abs(x[1]), reverse=True)\n",
    "# for f in feats:    \n",
    "#     print(f)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Linear Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-0.82404923647 [-1.1160617   0.01449197 -1.9734492  -0.05134546 -0.99388179]\n"
     ]
    }
   ],
   "source": [
    "lr = LinearRegression()\n",
    "from sklearn.model_selection import cross_val_score, KFold\n",
    "shuffle = KFold(n_splits=5, shuffle=True, random_state=0)\n",
    "scores = cross_val_score(lr, X, y, cv=shuffle)\n",
    "print(scores.mean(), scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.593674959591\n",
      "('aux_loss_grad_30', -116872976539522.78)\n",
      "('aux_loss_grad_70', -105356520782891.58)\n",
      "('aux_loss_grad_10', -74092294790496.25)\n",
      "('aux_loss_grad_20', 69294732384059.008)\n",
      "('aux_loss_grad_50', -67044131966574.938)\n",
      "('aux_curve_steepness0', -64936623283893.867)\n",
      "('aux_length', -56634101694847.766)\n",
      "('aux_tokens/types', 53616824301639.812)\n",
      "('aux_curve_steepness1', 44067220869800.133)\n",
      "('aux_js_div', 36442805318538.781)\n",
      "('aux_n_labels', -36385652100397.688)\n",
      "('aux_nuc_norm', 16539696322286.0)\n",
      "('main/aux_tokens/types', -49.04296875)\n",
      "('main/aux_length', 27.43359375)\n",
      "('main_curve_steepness1', 14.8203125)\n",
      "('main_loss_grad_50', 13.8642578125)\n",
      "('main/aux_curve_steepness1', -13.0078125)\n",
      "('main/aux_curve_steepness0', 12.671875)\n",
      "('main_js_div', 10.34375)\n",
      "('main_length', -9.998046875)\n",
      "('main/aux_loss_grad_20', 9.3271484375)\n",
      "('main_tokens/types', -6.0556640625)\n",
      "('main/aux_loss_grad_30', -5.748046875)\n",
      "('main_loss_grad_70', 5.2314453125)\n",
      "('main/aux_nuc_norm', 4.62109375)\n",
      "('main/aux_js_div', 3.90234375)\n",
      "('main/aux_loss_grad_50', 3.84326171875)\n",
      "('main_loss_grad_20', -3.556640625)\n",
      "('main_loss_grad_30', -3.546875)\n",
      "('main_curve_steepness0', -3.3828125)\n",
      "('main/aux_loss_grad_10', 3.078125)\n",
      "('main_loss_grad_10', 2.9609375)\n",
      "('main/aux_loss_grad_70', 2.91796875)\n",
      "('main_n_labels', -2.48046875)\n",
      "('main/aux_n_labels', 1.8828125)\n",
      "('main_nuc_norm', -1.7890625)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# X_train, X_test, y_train, y_test = train_test_split(xs, ys, test_size=0.2, random_state=42)\n",
    "lr = LinearRegression()\n",
    "\n",
    "lr.fit(X, y)\n",
    "print(lr.score(X, y))\n",
    "# print(lr.score(X_train, y_train))\n",
    "feats = sorted([(f, lr.coef_[i]) for i, f in enumerate(v.feature_names_)], key=lambda x: abs(x[1]), reverse=True)\n",
    "for f in feats:    \n",
    "    print(f)"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [Root]",
   "language": "python",
   "name": "Python [Root]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}