26-NYTimes-Data-Analysis.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"private_outputs":true,"provenance":[],"authorship_tag":"ABX9TyNXLzxlXHG9KRTdUSP0earF"},"kernelspec":{"name":"python3","display_name":"Python 3"},"language_info":{"name":"python"}},"cells":[{"cell_type":"code","execution_count":null,"metadata":{"id":"oKdLlaZ_fFTV"},"outputs":[],"source":["from google.colab import drive\n","drive.mount('/content/gdrive', force_remount=True)"]},{"cell_type":"code","source":["%cd /content/gdrive/MyDrive/DL/Facebook/fbscraper/nytimes"],"metadata":{"id":"yHQiFKLkfQ_U"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["import numpy as np\n","import pandas as pd\n","import matplotlib\n","from   matplotlib import pyplot as plt\n","import seaborn as sns"],"metadata":{"id":"sTn4bscWIfMV"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["sns.set(style='whitegrid')\n","# matplotlib.rcParams['figure.dpi'] = 120\n","matplotlib.rcParams['font.size'] = 18\n","# matplotlib.rcParams['figure.figsize'] = (10, 5)\n","matplotlib.rcParams['pdf.fonttype'] = 42\n","matplotlib.rcParams['ps.fonttype'] = 42"],"metadata":{"id":"lPMZqTSPInGn"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["import pickle as pkl\n","import pandas as pd \n","import numpy as np\n","import nltk\n","import re\n","import matplotlib.pyplot as plt\n","nltk.download('punkt') # For tokenizers\n","from nltk.tokenize import TweetTokenizer"],"metadata":{"id":"KgYn5syufktc"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["posts = []\n","\n","with open(\"nytimes.pkl\", \"rb\") as f:\n","    try:\n","        while True:\n","            posts.append(pkl.load(f))\n","    except EOFError:\n","        pass"],"metadata":{"id":"zywoCT6Rfusp"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["texts = [] # {text, authorName, authorURL, time, type}\n","\n","for post in posts:\n","    texts.append(dict(\n","        text=post[\"text\"],\n","        authorName=post[\"author\"][\"name\"],\n","        authorURL=post[\"author\"][\"url\"],\n","        time_=post[\"time\"],\n","        type_=\"post\"\n","    ))\n","    for comment in post[\"comments\"]:\n","        texts.append(dict(\n","            text=comment[\"text\"],\n","            authorName=comment[\"author\"][\"name\"],\n","            authorURL=comment[\"author\"][\"url\"],\n","            time_=post[\"time\"],\n","            type_=\"comment\"\n","        ))\n","        for reply in comment[\"replies\"]:\n","            texts.append(dict(\n","                text=reply[\"text\"],\n","                authorName=reply[\"author\"][\"name\"],\n","                authorURL=reply[\"author\"][\"url\"],\n","                time_=post[\"time\"],\n","                type_=\"reply\"\n","            ))"],"metadata":{"id":"KK680Qfjf9oG"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["len(texts)"],"metadata":{"id":"X2irDWSIu5KN"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["df = pd.DataFrame(texts)"],"metadata":{"id":"DFYLKGhlgAHC"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["df"],"metadata":{"id":"BZfxyNjgvbat"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["# Pre-processing text\n","\n","tknz = TweetTokenizer()\n","\n","def cleanText(text):\n","    text = text.lower()\n","    text = re.sub(r\"http\\S+\", \"\", text)\n","    text = re.sub(r\"www\\S+\", \"\", text)\n","    text = re.sub(\"-\", \" \", text)\n","    text = re.sub(\"\\s+\", \" \", text)\n","    text = re.sub(\"\\u2018\", \"X\", text) \n","    text = re.sub(\"\\u2019\", \"X\", text) \n","    text = re.sub(\"\\'\", \"X\", text) \n","    wordTokens_ = tknz.tokenize(text)\n","    wordTokens = list()\n","    for x in wordTokens_:\n","        x = ''.join([v for v in x if v.isalnum() or v == ' '])\n","        if len(x) > 0 and x != 'X':\n","            x = x.replace('X', '\\'')\n","            wordTokens.append(x)\n","    return ' '.join(wordTokens)"],"metadata":{"id":"jGx0Iao6jqyi"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["df[\"processedText\"] = df[\"text\"].apply(cleanText)"],"metadata":{"id":"dpqy0dPij8PF"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["df.to_csv(\"nytimes.csv\")"],"metadata":{"id":"r6XqOVqzlSAi"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["nComments = []\n","nReplies = []\n","\n","for post in posts:\n","    nComments.append(len(post[\"comments\"]))\n","    for comment in post[\"comments\"]:\n","        nReplies.append(len(comment[\"replies\"]))"],"metadata":{"id":"u7LOZ4PClmPi"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["nComments = np.array(nComments)\n","nReplies = np.array(nReplies)"],"metadata":{"id":"K4ifaBuvmxtB"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["nCommentsAvg = np.average(nComments)\n","nCommentsMed = np.median(nComments)\n","nRepliesAvg = np.average(nReplies)\n","nRepliesMed = np.median(nReplies)\n","print(nCommentsAvg, nCommentsMed, nRepliesAvg, nRepliesMed)"],"metadata":{"id":"0GEQCnHum3ST"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["authorName = df['authorName'].tolist()"],"metadata":{"id":"yQOkDlurnzVP"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["authorNameDist = dict()\n","for name in authorName:\n","    try:\n","        authorNameDist[name] += 1\n","    except:\n","        authorNameDist[name] = 1"],"metadata":{"id":"zFeoP9fHydYF"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["authorNameDistSorted = []\n","commits = []\n","for k, v in authorNameDist.items():\n","    authorNameDistSorted.append((v, k))\n","    commits.append(v)\n","authorNameDistSorted = sorted(authorNameDistSorted, reverse=True)\n","for i, name in enumerate(authorNameDistSorted[:10]):\n","    print(name)"],"metadata":{"id":"KM3VHjuC0HLC"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["Model Accuracy Precision (AH) Recall (AH) F1 (AH) Precision (None) Recall (None) F1 (None) Macro F1\n","BERT (10%) 0.721 0.796 0.594 0.680 0.676 0.848 0.752 0.716\n","BERT (20%) 0.785 0.786 0.782 0.784 0.783 0.787 0.785 0.784\n","BERT (30%) 0.804 0.812 0.790 0.801 0.796 0.817 0.806 0.803\n","BERT (40%) 0.811 0.794 0.840 0.816 0.830 0.782 0.805 0.811\n","BERT (50%) 0.824 0.801 0.860 0.829 0.848 0.787 0.817 0.823\n","BERT (60%) 0.831 0.818 0.850 0.834 0.844 0.811 0.827 0.830\n","BERT (70%) 0.836 0.835 0.838 0.836 0.837 0.834 0.836 0.836\n","BERT (80%) 0.832 0.838 0.822 0.830 0.825 0.841 0.833 0.831\n","BERT (90%) 0.837 0.845 0.826 0.835 0.830 0.848 0.839 0.837\n","BERT (100%) 0.839 0.837 0.840 0.839 0.840 0.837 0.838 0.838\n","GAN-BERT (10%) 0.777 0.767 0.796 0.781 0.788 0.758 0.773 0.777\n","GAN-BERT (20%) 0.792 0.802 0.775 0.788 0.782 0.809 0.795 0.792\n","GAN-BERT (30%) 0.798 0.796 0.802 0.799 0.800 0.794 0.797 0.798\n","GAN-BERT (40%) 0.798 0.788 0.815 0.801 0.808 0.781 0.794 0.798\n","GAN-BERT (50%) 0.807 0.813 0.797 0.805 0.801 0.817 0.809 0.807\n","GAN-BERT (60%) 0.813 0.818 0.804 0.811 0.807 0.821 0.814 0.812\n","GAN-BERT (70%) 0.814 0.807 0.824 0.815 0.820 0.803 0.812 0.814\n","GAN-BERT (80%) 0.820 0.825 0.811 0.818 0.814 0.828 0.821 0.820\n","GAN-BERT (90%) 0.818 0.805 0.839 0.822 0.831 0.797 0.814 0.805\n","GAN-BERT (100%) 0.839 0.837 0.840 0.839 0.840 0.837 0.838 0.838"],"metadata":{"id":"CZD-KIMQoEmU"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["x = [0.02, 0.04, 0.06, 0.08, 0.1, 0.12, 0.14, 0.16, 0.18, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]\n","\n","y1 = [0.597, 0.514, 0.739, 0.728, 0.716, 0.751, 0.745, 0.756, 0.772, 0.784, 0.803, 0.811, 0.823, 0.830, 0.836, 0.831, 0.837, 0.838] #bert\n","\n","y2 = [0.677, 0.745, 0.739, 0.752, 0.766, 0.782, 0.777, 0.792, 0.789, 0.792, 0.798, 0.798, 0.807, 0.812, 0.814, 0.820, 0.805, 0.838] #ganbert\n"],"metadata":{"id":"-U-3MuvGnhW3"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["data = {\n","    'fraction': [],\n","    'macro-f1': [],\n","    'model': []\n","}\n","\n","for i, j in zip(x, y1):\n","    data['fraction'].append(i)\n","    data['macro-f1'].append(j)\n","    data['model'].append('bert')\n","\n","for i, j in zip(x, y2):\n","    data['fraction'].append(i)\n","    data['macro-f1'].append(j)\n","    data['model'].append('gan-bert')\n","\n","fig, ax = plt.subplots()\n","sns.lineplot(data=data, x='fraction', y='macro-f1', hue='model', ax=ax)\n","ax.set_xlabel('fraction of labeled instances in training set')\n","ax.set_ylabel('macro-F1 score')\n","\n","fig.savefig('bert_vs_ganbert.pdf', bbox_inches='tight')\n"],"metadata":{"id":"Oh8p9QYdIxCz"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":["plt.xlabel('fraction of labeled instances in training set')\n","plt.ylabel('macro-F1 score')\n","plt.plot(x, y1, 'o-', label='BERT')\n","plt.plot(x, y2, 'o-', label='GAN-BERT')\n","plt.grid()\n","plt.legend()\n","plt.savefig('bert_vs_ganbert.eps', format='eps')\n","#plt.savefig(save_as)\n","#plt.close()"],"metadata":{"id":"5pSVI0TB09Uz"},"execution_count":null,"outputs":[]},{"cell_type":"code","source":[],"metadata":{"id":"89vQTHAFpkbR"},"execution_count":null,"outputs":[]}]}