update code of cross-validation over various labeled data size

code/.ipynb_checkpoints/EM_NB_text_classification_v4-checkpoint.ipynb

@@ -34,7 +34,8 @@
     "from copy import deepcopy\n",
     "from wordcloud import WordCloud \n",
     "from Semi_EM_NB import Semi_EM_MultinomialNB\n",
-    "\n",
+    "from os import path\n",
+    "from PIL import Image\n",
     "\n",
     "from nltk.corpus import stopwords\n",
     "from nltk.tokenize import RegexpTokenizer\n",
@@ -110,11 +111,13 @@
     "    feature_names = np.asarray(vectorizer.get_feature_names())\n",
     "    nrows, ncols = 5, 4\n",
     "    fig, axes = plt.subplots(figsize=(50, 40), nrows=nrows, ncols=ncols)\n",
+    "    d = path.dirname(__file__)\n",
+    "    circle_mask = np.array(Image.open(path.join(d, \"circle.png\")))\n",
     "    for i, category in enumerate(categories):\n",
     "        topK = np.argsort(classifier.coef_[i])[-K:]\n",
     "        text = \" \".join(feature_names[topK])\n",
     "        print(\"%s: %s\" % (category, text))\n",
-    "        wordcloud = WordCloud().generate(text)\n",
+    "        wordcloud = WordCloud(background_color=\"white\", mask=circle_mask,).generate(text)\n",
     "        axes[i//ncols, i%ncols].imshow(wordcloud, cmap=plt.cm.cool_r, interpolation='bilinear')\n",
     "        axes[i//ncols, i%ncols].axis(\"off\")\n",
     "        axes[i//ncols, i%ncols].set_title(category, fontweight=\"bold\", size=24)\n",