retweets.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analyze users by inspecting retweet counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import re\n",
    "from IPython.display import clear_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATADIR = \"/home/erikt/projects/puregome/data/retweets/\"\n",
    "RIGHT = \"geertwilderspvv thierrybaudet PvanHouwelingen fvdemocratie hansvantelling rblommestijn Kees71234\".split(\" \")\n",
    "DUTCHPRESS = \"telegraaf NOS RTLnieuws NUnl ADnl nrc volkskrant trouw parool AT5 NPORadio1 Nieuwsuur op1npo EenVandaag NPORadio1\".split(\" \")\n",
    "FLEMISHPRESS = \"HLN_BE demorgen destandaard vrtnws Nieuwsblad_be deafspraaktv VTMNIEUWS\".split(\" \")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def squeal(text=None):\n",
    "    clear_output(wait=True)\n",
    "    if not text is None: print(text)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Read the retweet data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20211231-23.out.gz\n"
     ]
    }
   ],
   "source": [
    "file_names = os.listdir(DATADIR)\n",
    "df_total = pd.DataFrame([])\n",
    "for file_name in sorted(file_names):\n",
    "    if re.search(r\"^2\", file_name):\n",
    "        df = pd.read_csv(DATADIR+file_name, sep=\",\", compression=\"gzip\").loc[:, [\"user\", \"retweeted_user\"]]\n",
    "        df_total = pd.concat([df_total, df])\n",
    "        squeal(file_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7981429"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_total)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Create a lists of retweeted users per retweeter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "groups = df_total.groupby([\"user\", \"retweeted_user\"]).groups"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Check how often each retweeter retweets users of the three main groups "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4457571/4457571\n"
     ]
    }
   ],
   "source": [
    "profiles = {}\n",
    "\n",
    "\n",
    "def add(user, group, count):\n",
    "    if not user in profiles: \n",
    "        profiles[user] = { RIGHT[0]: 0, DUTCHPRESS[0]: 0, FLEMISHPRESS[0]: 0}\n",
    "    profiles[user][group[0]] += count\n",
    "\n",
    "        \n",
    "counter = 0\n",
    "for key in groups.keys():\n",
    "    counter += 1\n",
    "    if counter % 10000 == 0: \n",
    "        squeal(str(counter)+\"/\"+str(len(groups)))\n",
    "    retweetedUser = key[1]\n",
    "    retweeter = key[0]\n",
    "    if retweetedUser in RIGHT: \n",
    "        add(retweeter, RIGHT, len(groups[key]))\n",
    "    elif retweetedUser in DUTCHPRESS: \n",
    "        add(retweeter, DUTCHPRESS, len(groups[key]))\n",
    "    elif retweetedUser in FLEMISHPRESS: \n",
    "        add(retweeter, FLEMISHPRESS, len(groups[key]))\n",
    "squeal(str(counter)+\"/\"+str(len(groups)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "{ key: profiles[key] for key in profiles if profiles[key][\"geertwilderspvv\"] >= 800 }"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Get users with retweets of only one group"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "MIN = 5\n",
    "usersPerGroup = {RIGHT[0]:[],DUTCHPRESS[0]:[],FLEMISHPRESS[0]:[]}\n",
    "for user in profiles:\n",
    "    if profiles[user][RIGHT[0]] > MIN and profiles[user][DUTCHPRESS[0]] == 0 and profiles[user][FLEMISHPRESS[0]] == 0:\n",
    "        usersPerGroup[RIGHT[0]].append(user)\n",
    "    if profiles[user][RIGHT[0]] == 0 and profiles[user][DUTCHPRESS[0]] > MIN and profiles[user][FLEMISHPRESS[0]] == 0:\n",
    "        usersPerGroup[DUTCHPRESS[0]].append(user)\n",
    "    if profiles[user][RIGHT[0]] == 0 and profiles[user][DUTCHPRESS[0]] == 0 and profiles[user][FLEMISHPRESS[0]] > MIN:\n",
    "        usersPerGroup[FLEMISHPRESS[0]].append(user)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4765, 987, 111)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(usersPerGroup[RIGHT[0]]),len(usersPerGroup[DUTCHPRESS[0]]),len(usersPerGroup[FLEMISHPRESS[0]])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Inspect favorite users per group"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_selected = df_total.loc[df_total[\"user\"].isin(usersPerGroup[RIGHT[0]])]\n",
    "df_selected[\"retweeted_user\"].value_counts()[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_selected = df_total.loc[df_total[\"user\"].isin(usersPerGroup[DUTCHPRESS[0]])]\n",
    "df_selected[\"retweeted_user\"].value_counts()[:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_selected = df_total.loc[df_total[\"user\"].isin(usersPerGroup[FLEMISHPRESS[0]])]\n",
    "df_selected[\"retweeted_user\"].value_counts()[:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Principal component analysis\n",
    "\n",
    "Source: https://nickmccullum.com/python-machine-learning/principal-component-analysis-python/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import math\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.decomposition import PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "COLORS = { \"RIGHT\": \"y\", \"DUTCHPRESS\": \"g\", \"FLEMISHPRESS\": \"b\", \"OTHER\": \"k\" }\n",
    "INVERTED_COLORS = { COLORS[x]: x for x in COLORS }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_scaler = StandardScaler()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dict = {}\n",
    "retweeted_counts = {}\n",
    "user_colors = []\n",
    "\n",
    "for key_pair in groups:\n",
    "    user, retweeted_user = key_pair\n",
    "    if retweeted_user not in retweeted_counts:\n",
    "        retweeted_counts[retweeted_user] = 0\n",
    "    retweeted_counts[retweeted_user] += len(groups[key_pair])\n",
    "\n",
    "\n",
    "for key_pair in groups:\n",
    "    user, retweeted_user = key_pair\n",
    "    if retweeted_counts[retweeted_user] > 10000:\n",
    "        if user not in data_dict:\n",
    "            data_dict[user] = {}\n",
    "        data_dict[user][retweeted_user] = len(groups[key_pair])\n",
    "\n",
    "        \n",
    "for user in data_dict:\n",
    "    for retweeted_user in data_dict[user]:\n",
    "        data_dict[user][retweeted_user] = math.log10(data_dict[user][retweeted_user])\n",
    "        \n",
    "        \n",
    "for user in data_dict:\n",
    "    if user in usersPerGroup[RIGHT[0]]:\n",
    "        user_colors.append(COLORS[\"RIGHT\"])\n",
    "    elif user in usersPerGroup[DUTCHPRESS[0]]:\n",
    "        user_colors.append(COLORS[\"DUTCHPRESS\"])\n",
    "    elif user in usersPerGroup[FLEMISHPRESS[0]]:\n",
    "        user_colors.append(COLORS[\"FLEMISHPRESS\"])\n",
    "    else:\n",
    "        user_colors.append(COLORS[\"OTHER\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_df = pd.DataFrame([data_dict[user] for user in data_dict]).fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(65204, 82)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data_df), len(data_df.loc[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StandardScaler()"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_scaler.fit(data_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaled_data_frame = data_scaler.transform(data_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "pca = PCA(n_components = 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PCA(n_components=4)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca.fit(scaled_data_frame)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_pca = pca.transform(scaled_data_frame)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "for color in [\"k\", \"y\", \"g\", \"b\"]:\n",
    "    x = [data_pca[i][0] for i in range(0, len(user_colors)) if user_colors[i] == color]\n",
    "    y = [data_pca[i][3] for i in range(0, len(user_colors)) if user_colors[i] == color]\n",
    "    c = [user_colors[i] for i in range(0, len(user_colors)) if user_colors[i] == color]\n",
    "    ax.scatter(x, y, c=c, alpha=0.3, label=INVERTED_COLORS[color])\n",
    "handles, labels = ax.get_legend_handles_labels()\n",
    "handles.reverse()\n",
    "labels.reverse()\n",
    "ax.legend(handles, labels)\n",
    "plt.title(\"Dutch-speaking Twitter users by retweet behavior\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "users = list(data_dict.keys())\n",
    "for i in range(0, len(users)):\n",
    "    if data_pca[i][0] > 80:\n",
    "        print(f\"{int(data_pca[i][0]):3} {int(data_pca[i][3]):2} {user_colors[i]} {users[i]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(0, len(users)):\n",
    "    if data_pca[i][3] > 23:\n",
    "        print(f\"{int(data_pca[i][0]):2} {int(data_pca[i][3]):2} {user_colors[i]} {users[i]}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Check for other interesting retweeted users per group (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "retweetedUsers = []\n",
    "counter = 0\n",
    "for user in usersPerGroup[RIGHT[0]]:\n",
    "    counter += 1\n",
    "    squeal(str(counter)+\"/\"+str(len(usersPerGroup[RIGHT[0]])))\n",
    "    retweetedUsers.extend([ x[1] for x in groups if x[0] == user ])\n",
    "retweetedUsersDict  = pd.DataFrame(retweetedUsers).groupby(0).groups"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "list({key:len(retweetedUsersDict[key]) for key in sorted(retweetedUsersDict.keys(),\n",
    "                                                         key=lambda k:len(retweetedUsersDict[k]),reverse=True)}.items())[0:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "list({key:len(retweetedUsersDict[key]) for key in sorted(retweetedUsersDict.keys(),\n",
    "                                                         key=lambda k:len(retweetedUsersDict[k]),reverse=True)}.items())[0:20]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "list({key:len(retweetedUsersDict[key]) for key in sorted(retweetedUsersDict.keys(),\n",
    "                                                         key=lambda k:len(retweetedUsersDict[k]),reverse=True)}.items())[0:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get tweet texts of selected users (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATADIRTEXT = \"/home/erikt/projects/puregome/data/text/\"\n",
    "FILEPATTERN = \"2020060[1-7]\"\n",
    "USER = \"user\"\n",
    "TEXT = \"text\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "selectedUsers = usersPerGroup[RIGHT[0]]+usersPerGroup[DUTCHPRESS[0]]+usersPerGroup[FLEMISHPRESS[0]]\n",
    "files = sorted(os.listdir(DATADIRTEXT))\n",
    "tweetTexts = {}\n",
    "for inFileName in files:\n",
    "    if re.search(FILEPATTERN,inFileName):\n",
    "        squeal(inFileName)\n",
    "        df = pd.read_csv(DATADIRTEXT+inFileName)\n",
    "        for i in range(0,len(df)):\n",
    "            user = df.iloc[i][USER]\n",
    "            if df.iloc[i][USER] in selectedUsers:\n",
    "                if not user in tweetTexts: tweetTexts[user] = df.iloc[i][TEXT]\n",
    "                else: tweetTexts[user] += \" \"+df.iloc[i][TEXT]\n",
    "tweetTexts"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tokenize tweet texts (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nltk.tokenize import TweetTokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def removeKeywords(text):\n",
    "    for keyword in RIGHT+DUTCHPRESS+FLEMISHPRESS:\n",
    "        text = re.sub(r\"\\b\"+keyword.lower()+r\"\\b\",\" \",text)\n",
    "    return(text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizedTweetTexts = {}\n",
    "tokenizer = TweetTokenizer()\n",
    "for user in tweetTexts.keys():\n",
    "    tokenizedTweetTexts[user] = \" \".join(tokenizer.tokenize(re.sub(r\"\\\\n\",\" \",removeKeywords(tweetTexts[user].lower()))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save tweet texts (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "OUTFILENAME = \"groups-train.txt\"\n",
    "LABELPREFIX = \"__label__\"\n",
    "RIGHTTEXT = \"RIGHT\"\n",
    "DUTCHPRESSTEXT = \"DUTCHPRESS\"\n",
    "FLEMISHPRESSTEXT = \"FLEMISHPRESS\"\n",
    "\n",
    "outFile = open(OUTFILENAME,\"w\")\n",
    "for user in tokenizedTweetTexts:\n",
    "    if user in usersPerGroup[RIGHT[0]]: label = RIGHTTEXT\n",
    "    elif user in usersPerGroup[DUTCHPRESS[0]]: label = DUTCHPRESSTEXT\n",
    "    elif user in usersPerGroup[FLEMISHPRESS[0]]: label = FLEMISHPRESSTEXT\n",
    "    else: sys.exit(\"cannot happen!\")\n",
    "    print(LABELPREFIX+label,tokenizedTweetTexts[user],file=outFile)\n",
    "outFile.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluate fasttext on data (optional)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import fasttext\n",
    "import os\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAINFILE = \"groups-train.txt\"\n",
    "LARGEINT = 99999999"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "fasttextData = []\n",
    "inFile = open(TRAINFILE,\"r\")\n",
    "for line in inFile: fasttextData.append(line.strip())\n",
    "inFile.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "DIM = 300\n",
    "EPOCH = 100\n",
    "LR = 0.1\n",
    "N = 10\n",
    "TRAIN = \"TRAIN\"+str(int(random.random()*LARGEINT))\n",
    "TEST = \"TEST\"+str(int(random.random()*LARGEINT))\n",
    "\n",
    "predictionCounts = []\n",
    "predictionLabels = []\n",
    "for fold in range(0,N):\n",
    "    squeal(\"starting fold \"+str(fold))\n",
    "    testStart = round(fold*len(fasttextData)/N)\n",
    "    testEnd = round((fold+1)*len(fasttextData)/N)\n",
    "    trainFile = open(TRAIN,\"w\")\n",
    "    testFile = open(TEST,\"w\")\n",
    "    testData = []\n",
    "    for i in range(0,len(fasttextData)):\n",
    "        if i < testStart or i >= testEnd: print(fasttextData[i],file=trainFile)\n",
    "        else: \n",
    "            print(fasttextData[i],file=testFile)\n",
    "            testData.append(fasttextData[i])\n",
    "    testFile.close()\n",
    "    trainFile.close()\n",
    "    model = fasttext.train_supervised(TRAIN,dim=DIM,epoch=EPOCH,lr=LR) # ,pretrainedVectors=PRETRAINEDDIR+WIKIFILENAME)\n",
    "    predictionCounts.append([*model.test(TEST)])\n",
    "    predictionLabels.append(model.predict(testData))\n",
    "    os.unlink(TRAIN)\n",
    "    os.unlink(TEST)\n",
    "clear_output(wait=True)\n",
    "print(\"finished\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "caseTotal = 0\n",
    "pTotal = 0\n",
    "rTotal = 0\n",
    "for i in range(0,len(predictionCounts)):\n",
    "    caseTotal += predictionCounts[i][0]\n",
    "    pTotal += predictionCounts[i][0]*predictionCounts[i][1]\n",
    "    rTotal += predictionCounts[i][0]*predictionCounts[i][2]\n",
    "print(\"cases: {0}; precision: {1}; recall: {2}\".format(caseTotal,round(pTotal/caseTotal,3),round(rTotal/caseTotal,3)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scores: epoch=10 P=52; epoch=20 P=86; epoch=30 P=89; epoch=50 P=91; epoch=100 P=94.7; epoch=200 P=94.8; epoch=130 P=95.1; epoch=150 P=95.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "labelCountsGold = {}\n",
    "for i in range(0,len(fasttextData)):\n",
    "    label = fasttextData[i].split()[0]\n",
    "    if label in labelCountsGold: labelCountsGold[label] += 1\n",
    "    else: labelCountsGold[label] = 1\n",
    "labelCountsGold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "labelCountsPredicted = {}\n",
    "for i in range(0,len(predictionLabels)):\n",
    "    for label in predictionLabels[i][0]:\n",
    "        if label[0] in labelCountsPredicted: labelCountsPredicted[label[0]] += 1\n",
    "        else: labelCountsPredicted[label[0]] = 1\n",
    "labelCountsPredicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(0,len(fasttextData)):\n",
    "    labelGold = fasttextData[i].split()[0]\n",
    "    labelPredicted = predictionLabels[i][0]\n",
    "    print(labelGold,labelPredicted)\n",
    "    break"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Apply fasttext to other data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import csv\n",
    "import fasttext\n",
    "import os\n",
    "import re\n",
    "import pandas as pd\n",
    "from nltk.tokenize import TweetTokenizer\n",
    "from IPython.display import clear_output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAINFILE = \"groups-train.txt\"\n",
    "DIM = 300\n",
    "EPOCH = 100\n",
    "LR = 0.1\n",
    "RIGHTTEXT = \"RIGHT\"\n",
    "DUTCHPRESSTEXT = \"DUTCHPRESS\"\n",
    "FLEMISHPRESSTEXT = \"FLEMISHPRESS\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def squeal(text=None):\n",
    "    clear_output(wait=True)\n",
    "    if not text is None: print(text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = fasttext.train_supervised(TRAINFILE,dim=DIM,epoch=EPOCH,lr=LR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "TEXT = \"text\"\n",
    "USER = \"user\"\n",
    "TESTFILEPATTERN = \"2020061[6-9]|2020062[0-2]\"\n",
    "DATADIRTEXT = \"/home/erikt/projects/puregome/data/text/\"\n",
    "\n",
    "files = sorted(os.listdir(DATADIRTEXT))\n",
    "tokenizer = TweetTokenizer()\n",
    "userTexts = {}\n",
    "for inFileName in files:\n",
    "    if re.search(TESTFILEPATTERN,inFileName):\n",
    "        squeal(inFileName)\n",
    "        df = pd.read_csv(DATADIRTEXT+inFileName)\n",
    "        for i in range(0,len(df)): \n",
    "            user = df.iloc[i][USER]\n",
    "            text = df.iloc[i][TEXT]\n",
    "            if user in userTexts: userTexts[user] += \" \"+text\n",
    "            else: userTexts[user] = text\n",
    "tokenizedTexts = {}\n",
    "for user in userTexts:\n",
    "    tokenizedTexts[user] = \" \".join(tokenizer.tokenize(re.sub(r\"\\\\n\",\" \",userTexts[user].lower())))\n",
    "predictedLabels = model.predict([tokenizedTexts[user] for user in tokenizedTexts.keys()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "LABEL = \"label\"\n",
    "CONFIDENCE = \"confidence\"\n",
    "\n",
    "predictedLabelsDict = {}\n",
    "counter = 0\n",
    "for user in tokenizedTexts:\n",
    "    predictedLabelsDict[user] = { LABEL:predictedLabels[0][counter][0], CONFIDENCE:predictedLabels[1][counter][0] }\n",
    "    counter += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "MAXCOUNTER = 1\n",
    "\n",
    "counter = 0\n",
    "for user in sorted(tokenizedTexts.keys(),key=lambda k:len(tokenizedTexts[k]),reverse=True):\n",
    "    print(predictedLabelsDict[user][LABEL],end=\" \")\n",
    "    print(predictedLabelsDict[user][CONFIDENCE],end=\" \")\n",
    "    if not user in profiles: print(\"???\",end=\" \")\n",
    "    elif profiles[user][RIGHT[0]] > 0: print(RIGHTTEXT,end=\" \")\n",
    "    elif profiles[user][FLEMISHPRESS[0]] > 0: print(FLEMISHPRESSTEXT,end=\" \")\n",
    "    elif profiles[user][DUTCHPRESS[0]] > 0: print(DUTCHPRESSTEXT,end=\" \")\n",
    "    else: print(\"?!?\",end=\" \")\n",
    "    print(tokenizedTexts[user][0:500])\n",
    "    counter += 1\n",
    "    if counter >= MAXCOUNTER: break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "MAXCOUNTER = 5000\n",
    "DATAFILEOUT = \"user-groups.csv\"\n",
    "ID = \"ID\"\n",
    "\n",
    "counter = 0\n",
    "outFile = open(DATAFILEOUT,\"w\")\n",
    "csvwriter = csv.DictWriter(outFile,[ID,USER,TEXT])\n",
    "for user in sorted(tokenizedTexts.keys(),key=lambda k:len(tokenizedTexts[k]),reverse=True):\n",
    "    csvwriter.writerow({ID:user,USER:user,TEXT:tokenizedTexts[user][0:1000]})\n",
    "    counter += 1\n",
    "    if counter >= MAXCOUNTER: break\n",
    "outFile.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}