sklearn-text-example-5.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "    <style>\n",
       "    table.eli5-weights tr:hover {\n",
       "        filter: brightness(85%);\n",
       "    }\n",
       "</style>\n",
       "\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "        \n",
       "\n",
       "    \n",
       "\n",
       "        \n",
       "\n",
       "        \n",
       "    \n",
       "        \n",
       "        \n",
       "    \n",
       "        <p style=\"margin-bottom: 0.5em; margin-top: 0em\">\n",
       "            <b>\n",
       "    \n",
       "        y=alt.atheism\n",
       "    \n",
       "</b>\n",
       "\n",
       "    \n",
       "    (score <b>-0.226</b>)\n",
       "\n",
       "top features\n",
       "        </p>\n",
       "    \n",
       "    <table class=\"eli5-weights\"\n",
       "           style=\"border-collapse: collapse; border: none; margin-top: 0em; table-layout: auto; margin-bottom: 2em;\">\n",
       "        <thead>\n",
       "        <tr style=\"border: none;\">\n",
       "            \n",
       "                <th style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\" title=\"Feature contribution already accounts for the feature value (for linear models, contribution = weight * feature value), and the sum of feature contributions is equal to the score or, for some classifiers, to the probability. Feature values are shown if &quot;show_feature_values&quot; is True.\">\n",
       "                    Contribution<sup>?</sup>\n",
       "                </th>\n",
       "            \n",
       "            <th style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">Feature</th>\n",
       "            \n",
       "        </tr>\n",
       "        </thead>\n",
       "        <tbody>\n",
       "        \n",
       "            <tr style=\"background-color: hsl(120, 100.00%, 83.95%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        +0.611\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        Highlighted in text (sum)\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "        \n",
       "\n",
       "        \n",
       "        \n",
       "            <tr style=\"background-color: hsl(0, 100.00%, 80.00%); border: none;\">\n",
       "    <td style=\"padding: 0 1em 0 0.5em; text-align: right; border: none;\">\n",
       "        -0.837\n",
       "    </td>\n",
       "    <td style=\"padding: 0 0.5em 0 0.5em; text-align: left; border: none;\">\n",
       "        &lt;BIAS&gt;\n",
       "    </td>\n",
       "    \n",
       "</tr>\n",
       "        \n",
       "\n",
       "        </tbody>\n",
       "    </table>\n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n",
       "    <p style=\"margin-bottom: 2.5em; margin-top:-0.5em;\">\n",
       "        <span style=\"opacity: 0.80\">\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 72.79%); opacity: 0.92\" title=\"0.045\">don</span><span style=\"opacity: 0.80\">'t be so </span><span style=\"background-color: hsl(0, 100.00%, 91.49%); opacity: 0.82\" title=\"-0.008\">sure</span><span style=\"opacity: 0.80\">.  </span><span style=\"background-color: hsl(0, 100.00%, 93.45%); opacity: 0.82\" title=\"-0.006\">look</span><span style=\"opacity: 0.80\"> what </span><span style=\"background-color: hsl(0, 100.00%, 91.11%); opacity: 0.82\" title=\"-0.009\">happened</span><span style=\"opacity: 0.80\"> to </span><span style=\"background-color: hsl(120, 100.00%, 93.19%); opacity: 0.82\" title=\"0.006\">japanese</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 84.14%); opacity: 0.85\" title=\"0.021\">citizens</span><span style=\"opacity: 0.80\"> in the us during\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 88.80%); opacity: 0.83\" title=\"0.013\">world</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 79.32%); opacity: 0.88\" title=\"0.030\">war</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 88.36%); opacity: 0.83\" title=\"0.013\">ii</span><span style=\"opacity: 0.80\">.  if you're </span><span style=\"background-color: hsl(120, 100.00%, 82.52%); opacity: 0.86\" title=\"0.024\">prepared</span><span style=\"opacity: 0.80\"> to </span><span style=\"background-color: hsl(120, 100.00%, 75.89%); opacity: 0.90\" title=\"0.038\">say</span><span style=\"opacity: 0.80\"> &quot;</span><span style=\"background-color: hsl(120, 100.00%, 93.52%); opacity: 0.81\" title=\"0.006\">let</span><span style=\"opacity: 0.80\">'s </span><span style=\"background-color: hsl(120, 100.00%, 91.64%); opacity: 0.82\" title=\"0.008\">round</span><span style=\"opacity: 0.80\"> these </span><span style=\"background-color: hsl(120, 100.00%, 75.89%); opacity: 0.90\" title=\"0.038\">people</span><span style=\"opacity: 0.80\"> up and\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 71.75%); opacity: 0.92\" title=\"0.047\">stick</span><span style=\"opacity: 0.80\"> them in a </span><span style=\"background-color: hsl(120, 100.00%, 87.78%); opacity: 0.84\" title=\"0.014\">concentration</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 91.58%); opacity: 0.82\" title=\"0.008\">camp</span><span style=\"opacity: 0.80\"> without </span><span style=\"background-color: hsl(120, 100.00%, 85.84%); opacity: 0.85\" title=\"0.018\">trial</span><span style=\"opacity: 0.80\">&quot;, it's only a </span><span style=\"background-color: hsl(0, 100.00%, 82.92%); opacity: 0.86\" title=\"-0.023\">short</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(0, 100.00%, 86.99%); opacity: 0.84\" title=\"-0.016\">step</span><span style=\"opacity: 0.80\"> to\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 86.11%); opacity: 0.84\" title=\"0.017\">gassing</span><span style=\"opacity: 0.80\"> them without </span><span style=\"background-color: hsl(120, 100.00%, 85.84%); opacity: 0.85\" title=\"0.018\">trial</span><span style=\"opacity: 0.80\">.  after all, it seems that the </span><span style=\"background-color: hsl(120, 100.00%, 88.88%); opacity: 0.83\" title=\"0.012\">nazis</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 92.75%); opacity: 0.82\" title=\"0.007\">originally</span><span style=\"opacity: 0.80\">\n",
       "only </span><span style=\"background-color: hsl(0, 100.00%, 94.10%); opacity: 0.81\" title=\"-0.005\">intended</span><span style=\"opacity: 0.80\"> to </span><span style=\"background-color: hsl(120, 100.00%, 86.11%); opacity: 0.84\" title=\"0.017\">imprison</span><span style=\"opacity: 0.80\"> the </span><span style=\"background-color: hsl(0, 100.00%, 96.10%); opacity: 0.81\" title=\"-0.003\">jews</span><span style=\"opacity: 0.80\">; the </span><span style=\"background-color: hsl(0, 100.00%, 91.35%); opacity: 0.82\" title=\"-0.009\">final</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(0, 100.00%, 94.18%); opacity: 0.81\" title=\"-0.005\">solution</span><span style=\"opacity: 0.80\"> was </span><span style=\"background-color: hsl(120, 100.00%, 86.11%); opacity: 0.84\" title=\"0.017\">dreamt</span><span style=\"opacity: 0.80\"> up </span><span style=\"background-color: hsl(120, 100.00%, 87.07%); opacity: 0.84\" title=\"0.015\">partly</span><span style=\"opacity: 0.80\">\n",
       "because they </span><span style=\"background-color: hsl(120, 100.00%, 72.37%); opacity: 0.92\" title=\"0.046\">couldn</span><span style=\"opacity: 0.80\">'t </span><span style=\"background-color: hsl(120, 100.00%, 87.08%); opacity: 0.84\" title=\"0.015\">afford</span><span style=\"opacity: 0.80\"> to </span><span style=\"background-color: hsl(0, 100.00%, 85.74%); opacity: 0.85\" title=\"-0.018\">run</span><span style=\"opacity: 0.80\"> the </span><span style=\"background-color: hsl(120, 100.00%, 87.38%); opacity: 0.84\" title=\"0.015\">camps</span><span style=\"opacity: 0.80\"> because of the </span><span style=\"background-color: hsl(120, 100.00%, 90.26%); opacity: 0.83\" title=\"0.010\">devastation</span><span style=\"opacity: 0.80\">\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 93.70%); opacity: 0.81\" title=\"0.006\">caused</span><span style=\"opacity: 0.80\"> by </span><span style=\"background-color: hsl(120, 100.00%, 86.11%); opacity: 0.84\" title=\"0.017\">goering</span><span style=\"opacity: 0.80\">'s </span><span style=\"background-color: hsl(120, 100.00%, 84.81%); opacity: 0.85\" title=\"0.019\">total</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 79.32%); opacity: 0.88\" title=\"0.030\">war</span><span style=\"opacity: 0.80\">.  those who </span><span style=\"background-color: hsl(120, 100.00%, 60.00%); opacity: 1.00\" title=\"0.077\">weren</span><span style=\"opacity: 0.80\">'t </span><span style=\"background-color: hsl(120, 100.00%, 86.11%); opacity: 0.84\" title=\"0.017\">gassed</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(0, 100.00%, 87.57%); opacity: 0.84\" title=\"-0.015\">generally</span><span style=\"opacity: 0.80\"> </span><span style=\"background-color: hsl(120, 100.00%, 80.00%); opacity: 0.87\" title=\"0.029\">died</span><span style=\"opacity: 0.80\"> of\n",
       "</span><span style=\"background-color: hsl(120, 100.00%, 87.08%); opacity: 0.84\" title=\"0.015\">malnutrition</span><span style=\"opacity: 0.80\"> or </span><span style=\"background-color: hsl(0, 100.00%, 97.64%); opacity: 0.80\" title=\"-0.001\">disease</span><span style=\"opacity: 0.80\">.\n",
       "</span>\n",
       "    </p>\n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "    \n",
       "\n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "Explanation(estimator=\"SGDClassifier(alpha=0.001, average=False, class_weight=None, epsilon=0.1,\\n       eta0=0.0, fit_intercept=True, l1_ratio=0.15,\\n       learning_rate='optimal', loss='hinge', n_iter=5, n_jobs=1,\\n       penalty='l2', power_t=0.5, random_state=42, shuffle=True, verbose=0,\\n       warm_start=False)\", description=None, error=None, method='linear model', is_regression=False, targets=[TargetExplanation(target='alt.atheism', feature_weights=FeatureWeights(pos=[FeatureWeight(feature='weren', weight=0.0773954100541661, std=None, value=0.14113251347628228), FeatureWeight(feature='war', weight=0.060331064235727434, std=None, value=0.25329243510498373), FeatureWeight(feature='stick', weight=0.047099798280833598, std=None, value=0.13878354649278488), FeatureWeight(feature='couldn', weight=0.045627494068862393, std=None, value=0.11665332745553016), FeatureWeight(feature='don', weight=0.044633578336684819, std=None, value=0.059008669069186982), FeatureWeight(feature='people', weight=0.03756098108116717, std=None, value=0.061605331599386762), FeatureWeight(feature='say', weight=0.037547593579566788, std=None, value=0.070035159108064279), FeatureWeight(feature='trial', weight=0.035103971266393565, std=None, value=0.29326399549283061), FeatureWeight(feature='died', weight=0.028746754338132854, std=None, value=0.1217005880387947), FeatureWeight(feature='prepared', weight=0.023715008933589188, std=None, value=0.13113962337932056), FeatureWeight(feature='citizens', weight=0.020645220994210334, std=None, value=0.14822258183331075), FeatureWeight(feature='total', weight=0.019411252142228472, std=None, value=0.12805491520907761), FeatureWeight(feature='imprison', weight=0.017081411509440175, std=None, value=0.1978808731083859), FeatureWeight(feature='gassing', weight=0.017081411509440175, std=None, value=0.1978808731083859), FeatureWeight(feature='goering', weight=0.017081411509440175, std=None, value=0.1978808731083859), FeatureWeight(feature='gassed', weight=0.017081411509440175, std=None, value=0.1978808731083859), FeatureWeight(feature='dreamt', weight=0.017081411509440175, std=None, value=0.1978808731083859), FeatureWeight(feature='partly', weight=0.015420739822327539, std=None, value=0.17080502455743396), FeatureWeight(feature='malnutrition', weight=0.015399766224814532, std=None, value=0.18788798301142415), FeatureWeight(feature='afford', weight=0.015399766224814532, std=None, value=0.18788798301142415), FeatureWeight(feature='camps', weight=0.014886086253111977, std=None, value=0.17080502455743396), FeatureWeight(feature='concentration', weight=0.014231245031190482, std=None, value=0.16700590406326113), FeatureWeight(feature='ii', weight=0.013267434659514979, std=None, value=0.13768801060300351), FeatureWeight(feature='world', weight=0.012561859446063313, std=None, value=0.087675101707034339), FeatureWeight(feature='nazis', weight=0.012429767025009412, std=None, value=0.17529843038426265), FeatureWeight(feature='devastation', weight=0.010285112920622628, std=None, value=0.18788798301142415), FeatureWeight(feature='camp', weight=0.0083609832490070066, std=None, value=0.15821547193027247), FeatureWeight(feature='round', weight=0.0082766677954305527, std=None, value=0.15174937359587662), FeatureWeight(feature='originally', weight=0.0067404117855649811, std=None, value=0.13197514405233782), FeatureWeight(feature='japanese', weight=0.0061648526753915909, std=None, value=0.16700590406326113), FeatureWeight(feature='let', weight=0.005741759401189735, std=None, value=0.091972128013194321), FeatureWeight(feature='caused', weight=0.0055198750668633599, std=None, value=0.11904797656414623)], neg=[FeatureWeight(feature='<BIAS>', weight=-0.83683508055376643, std=None, value=1.0), FeatureWeight(feature='short', weight=-0.022953012470607569, std=None, value=0.10922331578554058), FeatureWeight(feature='run', weight=-0.017726710033409473, std=None, value=0.10956325909850168), FeatureWeight(feature='step', weight=-0.015549760669647097, std=None, value=0.12284709705831905), FeatureWeight(feature='generally', weight=-0.014569310446517788, std=None, value=0.11447088475047355), FeatureWeight(feature='happened', weight=-0.0090320514326100703, std=None, value=0.11904797656414623), FeatureWeight(feature='final', weight=-0.0086863622374825557, std=None, value=0.12597095614933734), FeatureWeight(feature='sure', weight=-0.0084809410873312695, std=None, value=0.084017216553222904), FeatureWeight(feature='look', weight=-0.0058345855656802403, std=None, value=0.089577478904578248), FeatureWeight(feature='intended', weight=-0.0050267547796723272, std=None, value=0.11904797656414623), FeatureWeight(feature='solution', weight=-0.004931561536108256, std=None, value=0.13113962337932056), FeatureWeight(feature='jews', weight=-0.0027801630252263926, std=None, value=0.11620110376866166), FeatureWeight(feature='disease', weight=-0.0013547744822263942, std=None, value=0.10518421774345865)], pos_remaining=0, neg_remaining=0), proba=None, score=-0.22584955588060562, weighted_spans=WeightedSpans(docs_weighted_spans=[DocWeightedSpans(document='\\ndon\\'t be so sure.  look what happened to japanese citizens in the us during\\nworld war ii.  if you\\'re prepared to say \"let\\'s round these people up and\\nstick them in a concentration camp without trial\", it\\'s only a short step to\\ngassing them without trial.  after all, it seems that the nazis originally\\nonly intended to imprison the jews; the final solution was dreamt up partly\\nbecause they couldn\\'t afford to run the camps because of the devastation\\ncaused by goering\\'s total war.  those who weren\\'t gassed generally died of\\nmalnutrition or disease.\\n', spans=[('don', [(1, 4)], 0.044633578336684819), ('sure', [(13, 17)], -0.0084809410873312695), ('look', [(20, 24)], -0.0058345855656802403), ('happened', [(30, 38)], -0.0090320514326100703), ('japanese', [(42, 50)], 0.0061648526753915909), ('citizens', [(51, 59)], 0.020645220994210334), ('world', [(77, 82)], 0.012561859446063313), ('war', [(83, 86)], 0.060331064235727434), ('ii', [(87, 89)], 0.013267434659514979), ('prepared', [(102, 110)], 0.023715008933589188), ('say', [(114, 117)], 0.037547593579566788), ('let', [(119, 122)], 0.005741759401189735), ('round', [(125, 130)], 0.0082766677954305527), ('people', [(137, 143)], 0.03756098108116717), ('stick', [(151, 156)], 0.047099798280833598), ('concentration', [(167, 180)], 0.014231245031190482), ('camp', [(181, 185)], 0.0083609832490070066), ('trial', [(194, 199)], 0.035103971266393565), ('short', [(214, 219)], -0.022953012470607569), ('step', [(220, 224)], -0.015549760669647097), ('gassing', [(228, 235)], 0.017081411509440175), ('trial', [(249, 254)], 0.035103971266393565), ('nazis', [(286, 291)], 0.012429767025009412), ('originally', [(292, 302)], 0.0067404117855649811), ('intended', [(308, 316)], -0.0050267547796723272), ('imprison', [(320, 328)], 0.017081411509440175), ('jews', [(333, 337)], -0.0027801630252263926), ('final', [(343, 348)], -0.0086863622374825557), ('solution', [(349, 357)], -0.004931561536108256), ('dreamt', [(362, 368)], 0.017081411509440175), ('partly', [(372, 378)], 0.015420739822327539), ('couldn', [(392, 398)], 0.045627494068862393), ('afford', [(401, 407)], 0.015399766224814532), ('run', [(411, 414)], -0.017726710033409473), ('camps', [(419, 424)], 0.014886086253111977), ('devastation', [(440, 451)], 0.010285112920622628), ('caused', [(452, 458)], 0.0055198750668633599), ('goering', [(462, 469)], 0.017081411509440175), ('total', [(472, 477)], 0.019411252142228472), ('war', [(478, 481)], 0.060331064235727434), ('weren', [(494, 499)], 0.0773954100541661), ('gassed', [(502, 508)], 0.017081411509440175), ('generally', [(509, 518)], -0.014569310446517788), ('died', [(519, 523)], 0.028746754338132854), ('malnutrition', [(527, 539)], 0.015399766224814532), ('disease', [(543, 550)], -0.0013547744822263942)], preserve_density=False, vec_name=None)], other=FeatureWeights(pos=[FeatureWeight(feature=<FormattedFeatureName 'Highlighted in text (sum)'>, weight=0.61098552467316103, std=None, value=None)], neg=[FeatureWeight(feature='<BIAS>', weight=-0.83683508055376643, std=None, value=1.0)], pos_remaining=0, neg_remaining=0)))], feature_importances=None, decision_tree=None, highlight_spaces=None, transition_features=None)"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "doc = twenty_train.data[53]\n",
    "eli5.explain_prediction(clf, doc, vec=vec, top_targets=1,\n",
    "                        target_names=twenty_train.target_names)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}