example.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "822be3b0-db1e-4113-acad-e5854c359220",
   "metadata": {},
   "outputs": [],
   "source": [
    "import copy\n",
    "import random\n",
    "import math\n",
    "import time\n",
    "import pickle\n",
    "import pprint\n",
    "\n",
    "import tqdm\n",
    "\n",
    "import numpy\n",
    "import pandas\n",
    "\n",
    "import sklearn\n",
    "import sklearn.metrics\n",
    "import sklearn.svm\n",
    "\n",
    "import xgboost\n",
    "\n",
    "import matplotlib\n",
    "import matplotlib.pyplot\n",
    "import seaborn"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "6e85e311-1d71-4eef-9b27-1e9d1ba6fc6b",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Utility functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6992db6a",
   "metadata": {},
   "outputs": [],
   "source": [
    "def equal_pdf(size):\n",
    "    return numpy.ones((size,)) / size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "49dff043-df4c-4eae-8f4b-12759623db00",
   "metadata": {},
   "outputs": [],
   "source": [
    "def normalize_pdf(x):\n",
    "    x = x / x.sum()\n",
    "    x = numpy.nan_to_num(x, copy=True, nan=0.0, posinf=None, neginf=None)\n",
    "    if x.sum() == 0.:\n",
    "        x = equal_pdf(x.shape[0])\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "268e8234-1bac-4e02-97a5-4864d7555f4b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_k_accuracy_for_target(x, target):\n",
    "    x = pandas.Series(x).sort_values(ascending=False).reset_index()\n",
    "    return x[x['index']==target].index[0] + len(x[x[0]==x[x['index']==target].iloc[0][0]].index)/2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "9462572c-ebb8-4088-aa98-dde1af854aa5",
   "metadata": {},
   "outputs": [],
   "source": [
    "def target_pdf(size, target):\n",
    "    x = numpy.zeros((size,))\n",
    "    x[target] = 1\n",
    "    return x"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1b80f535-4cfd-4bfe-a3b9-f30085c25c4e",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Opinion pools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fb42bfc7-8340-4970-bec5-ac35bed9e836",
   "metadata": {},
   "outputs": [],
   "source": [
    "def majority_vote(X):\n",
    "    votes = X.argmax(axis=1)\n",
    "    vote = pandas.Series(votes).value_counts().sort_index().idxmax()\n",
    "    return target_pdf(X.shape[1], vote)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "2e15553c-38a6-4ec5-a168-a1cb1cc1276c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def linear_opinion_pool(X):\n",
    "    x = X.sum(axis=0)\n",
    "    return normalize_pdf(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1c1230a1-149b-4fae-96c4-724f7d77d814",
   "metadata": {},
   "outputs": [],
   "source": [
    "def logarithmic_opinion_pool(X):\n",
    "    x = X.prod(axis=0)\n",
    "    return normalize_pdf(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6b40c619-4193-499e-9d84-f19c973cdb41",
   "metadata": {},
   "outputs": [],
   "source": [
    "def paired_opinion_pool(X, pool1, pool2):\n",
    "    Y = []\n",
    "    for i in range(X.shape[0]):\n",
    "        for j in range(i+1, X.shape[0]):\n",
    "            Y.append(pool1(numpy.array([X[i, :], X[j, :]])))\n",
    "    Y = numpy.array(Y)\n",
    "    return pool2(Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "98f3e82f-d992-44ab-83aa-8edd25e66ae7",
   "metadata": {},
   "outputs": [],
   "source": [
    "class PairedOpinionPool:\n",
    "    def __init__(self, pool1, pool2):\n",
    "        self.pool1 = pool1\n",
    "        self.pool2 = pool2\n",
    "        \n",
    "    def __call__(self, X):\n",
    "        return paired_opinion_pool(X, self.pool1, self.pool2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ea98bf7c-739f-43f2-8dc6-05cd0317954f",
   "metadata": {},
   "outputs": [],
   "source": [
    "logarithmic_linear_opinion_pool = PairedOpinionPool(logarithmic_opinion_pool, linear_opinion_pool)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1d793e75-cff2-4b07-9eaf-d16dbcf8179d",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a5a2673d-8f74-437b-958c-f60b2a5d5b2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 100000\n",
    "EXPERIMENT_SIZE = 100000\n",
    "ROLLING_WINDOW_SIZE = EXPERIMENT_SIZE//10\n",
    "# 128\n",
    "ACTORS = 3\n",
    "# 8\n",
    "FEATURES = 3\n",
    "\n",
    "# 0.1\n",
    "BASE_ACTIVITY = 0.1\n",
    "BASE_STD = 1/ACTORS\n",
    "\n",
    "ACTIVITY_SHIFT_STD = 1e-2\n",
    "MEAN_SHIFT_STD = 1e-2\n",
    "STD_SHIFT_STD = 1e-2\n",
    "\n",
    "TRAIN_SIZE = 0.5\n",
    "\n",
    "PROB_FALSE_FLAG = 0.4\n",
    "\n",
    "MODELS = {\n",
    "    # 'Naive Bayes': sklearn.naive_bayes.GaussianNB(),\n",
    "    'Linear SVM': sklearn.svm.SVC(kernel='linear', probability=True),\n",
    "    # 'SVM with RBF': sklearn.svm.SVC(probability=True),\n",
    "    # 'MLP': sklearn.neural_network.MLPClassifier(early_stopping=True),\n",
    "    'XGBoost': xgboost.XGBClassifier(n_estimators=1000, n_jobs=1, objective='multi_softmax')\n",
    "}\n",
    "MODEL = MODELS['Linear SVM']\n",
    "\n",
    "FIGSIZE = (8, 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e03adfbf-83a2-4659-b0c9-896e5022736c",
   "metadata": {},
   "outputs": [],
   "source": [
    "pools = {\n",
    "    # 'Majority Vote': majority_vote, \n",
    "    'Linear Opinion Pool': linear_opinion_pool, \n",
    "    'Logarithmic Opinion Pool': logarithmic_opinion_pool,\n",
    "    'Pairing Aggregator': logarithmic_linear_opinion_pool,\n",
    "}"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "2d583518-00a4-4609-a71c-2d177594ed52",
   "metadata": {},
   "source": [
    "# Experiment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "9cf24d69",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_names = [f'x{i}' for i in range(FEATURES)]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6bbcf9c5",
   "metadata": {},
   "source": [
    "## Prepare dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ff0b6329",
   "metadata": {},
   "outputs": [],
   "source": [
    "columns = [\n",
    "    'Name',\n",
    "    'start',\n",
    "    'end',\n",
    "    'activity',\n",
    "]\n",
    "\n",
    "for i in range(FEATURES):\n",
    "    columns.append(f'x{i}_mean')\n",
    "    columns.append(f'x{i}_std')\n",
    "\n",
    "df_actors = pandas.DataFrame(columns=columns)\n",
    "\n",
    "for i in range(ACTORS):\n",
    "    start = random.uniform(0, EXPERIMENT_SIZE * (TRAIN_SIZE - 0.1))\n",
    "    end = random.uniform(EXPERIMENT_SIZE * (TRAIN_SIZE + 0.1), EXPERIMENT_SIZE)\n",
    "\n",
    "    values = [\n",
    "        'APT'+str(i),\n",
    "        start,\n",
    "        end,\n",
    "        random.uniform(0.0001, BASE_ACTIVITY),\n",
    "    ]\n",
    "    for _ in range(FEATURES):\n",
    "        values.append(random.uniform(-1, 1))\n",
    "        values.append(random.uniform(0, BASE_STD))\n",
    "\n",
    "    df_actors.loc[i] = pandas.Series(\n",
    "        values,\n",
    "        index=df_actors.columns,\n",
    "        name=i,\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "2f72d1ac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Name</th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>activity</th>\n",
       "      <th>x0_mean</th>\n",
       "      <th>x0_std</th>\n",
       "      <th>x1_mean</th>\n",
       "      <th>x1_std</th>\n",
       "      <th>x2_mean</th>\n",
       "      <th>x2_std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>34400.027497</td>\n",
       "      <td>99232.401041</td>\n",
       "      <td>0.05293</td>\n",
       "      <td>0.616601</td>\n",
       "      <td>0.127307</td>\n",
       "      <td>-0.995215</td>\n",
       "      <td>0.307068</td>\n",
       "      <td>-0.218063</td>\n",
       "      <td>0.045395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9243.730191</td>\n",
       "      <td>87094.169063</td>\n",
       "      <td>0.083828</td>\n",
       "      <td>0.518659</td>\n",
       "      <td>0.290513</td>\n",
       "      <td>-0.226985</td>\n",
       "      <td>0.270489</td>\n",
       "      <td>0.22577</td>\n",
       "      <td>0.209123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>APT2</td>\n",
       "      <td>35951.65766</td>\n",
       "      <td>81759.618103</td>\n",
       "      <td>0.053876</td>\n",
       "      <td>-0.12686</td>\n",
       "      <td>0.132722</td>\n",
       "      <td>-0.277634</td>\n",
       "      <td>0.267092</td>\n",
       "      <td>0.705371</td>\n",
       "      <td>0.114815</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Name         start           end  activity   x0_mean    x0_std   x1_mean  \\\n",
       "0  APT0  34400.027497  99232.401041   0.05293  0.616601  0.127307 -0.995215   \n",
       "1  APT1   9243.730191  87094.169063  0.083828  0.518659  0.290513 -0.226985   \n",
       "2  APT2   35951.65766  81759.618103  0.053876  -0.12686  0.132722 -0.277634   \n",
       "\n",
       "     x1_std   x2_mean    x2_std  \n",
       "0  0.307068 -0.218063  0.045395  \n",
       "1  0.270489   0.22577  0.209123  \n",
       "2  0.267092  0.705371  0.114815  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_actors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "d32a0d43",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 154/100000 [00:00<01:04, 1536.37it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 100000/100000 [01:46<00:00, 939.89it/s]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns = [\n",
    "    'Name',\n",
    "    'Timestep',\n",
    "] + feature_names\n",
    "\n",
    "df_incidents = pandas.DataFrame(columns=columns)\n",
    "\n",
    "incident_i = 0\n",
    "for i in tqdm.tqdm(range(EXPERIMENT_SIZE)):\n",
    "    for j in range(ACTORS):\n",
    "        if df_actors.iloc[j]['start'] <= i and df_actors.iloc[j]['end'] >= i and random.uniform(0, 1) < df_actors.iloc[j]['activity']:\n",
    "            feature_values = [\n",
    "                random.normalvariate(\n",
    "                    df_actors.iloc[j][f'x{k}_mean'], df_actors.iloc[j][f'x{k}_std']\n",
    "                )\n",
    "                for k in range(FEATURES)\n",
    "            ]\n",
    "\n",
    "            # If generating test data already, introduce false flags corresponding to a sample from another actor present in the training data\n",
    "            if i > EXPERIMENT_SIZE * TRAIN_SIZE:\n",
    "                # if len(df_incidents[(df_incidents['Name'] != df_actors.iloc[j]['Name'])].index) > 0:\n",
    "                blamed_actor = df_incidents[\n",
    "                    (df_incidents['Name'] != df_actors.iloc[j]['Name'])\n",
    "                    & (df_incidents['Timestep'] < EXPERIMENT_SIZE * TRAIN_SIZE)\n",
    "                ]['Name'].sample(1).iloc[0]\n",
    "                for k in range(FEATURES):\n",
    "                    if random.uniform(0, 1) < PROB_FALSE_FLAG:\n",
    "                        feature_values[k] = df_incidents[\n",
    "                            (df_incidents['Name'] == blamed_actor)\n",
    "                            & (df_incidents['Timestep'] < EXPERIMENT_SIZE * TRAIN_SIZE)\n",
    "                        ][f'x{k}'].sample(1).iloc[0]\n",
    "\n",
    "            values = [\n",
    "                df_actors.iloc[j]['Name'],\n",
    "                i,\n",
    "            ] + feature_values\n",
    "\n",
    "            series_index = str(i)+str(',')+str(j)\n",
    "\n",
    "            df_incidents.loc[series_index] = pandas.Series(\n",
    "                values,\n",
    "                index=df_incidents.columns,\n",
    "                name=series_index,\n",
    "            )\n",
    "            incident_i += 1\n",
    "\n",
    "    for k in range(FEATURES):\n",
    "        df_actors.loc[j, 'activity'] = min(\n",
    "            max(\n",
    "                df_actors.loc[j, 'activity'] +\n",
    "                random.normalvariate(0, ACTIVITY_SHIFT_STD),\n",
    "                0\n",
    "            ),\n",
    "            BASE_ACTIVITY\n",
    "        )\n",
    "\n",
    "        df_actors.loc[j, f'x{k}_mean'] = min(\n",
    "            max(\n",
    "                df_actors.loc[j, f'x{k}_mean'] +\n",
    "                random.normalvariate(0, MEAN_SHIFT_STD),\n",
    "                -1\n",
    "            ),\n",
    "            1\n",
    "        )\n",
    "\n",
    "        df_actors.loc[j, f'x{k}_std'] = min(\n",
    "            max(\n",
    "                df_actors.loc[j, f'x{k}_std'] +\n",
    "                random.normalvariate(0, STD_SHIFT_STD),\n",
    "                0\n",
    "            ),\n",
    "            BASE_STD\n",
    "        )\n",
    "\n",
    "    if i % 1000 == 0:\n",
    "        df_incidents = df_incidents.copy()\n",
    "0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "84fa3422",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Name</th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>activity</th>\n",
       "      <th>x0_mean</th>\n",
       "      <th>x0_std</th>\n",
       "      <th>x1_mean</th>\n",
       "      <th>x1_std</th>\n",
       "      <th>x2_mean</th>\n",
       "      <th>x2_std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>34400.027497</td>\n",
       "      <td>99232.401041</td>\n",
       "      <td>0.05293</td>\n",
       "      <td>0.616601</td>\n",
       "      <td>0.127307</td>\n",
       "      <td>-0.995215</td>\n",
       "      <td>0.307068</td>\n",
       "      <td>-0.218063</td>\n",
       "      <td>0.045395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9243.730191</td>\n",
       "      <td>87094.169063</td>\n",
       "      <td>0.083828</td>\n",
       "      <td>0.518659</td>\n",
       "      <td>0.290513</td>\n",
       "      <td>-0.226985</td>\n",
       "      <td>0.270489</td>\n",
       "      <td>0.22577</td>\n",
       "      <td>0.209123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>APT2</td>\n",
       "      <td>35951.65766</td>\n",
       "      <td>81759.618103</td>\n",
       "      <td>0.023797</td>\n",
       "      <td>-0.759649</td>\n",
       "      <td>0.282624</td>\n",
       "      <td>0.62864</td>\n",
       "      <td>0.00722</td>\n",
       "      <td>0.049739</td>\n",
       "      <td>0.215013</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Name         start           end  activity   x0_mean    x0_std   x1_mean  \\\n",
       "0  APT0  34400.027497  99232.401041   0.05293  0.616601  0.127307 -0.995215   \n",
       "1  APT1   9243.730191  87094.169063  0.083828  0.518659  0.290513 -0.226985   \n",
       "2  APT2   35951.65766  81759.618103  0.023797 -0.759649  0.282624   0.62864   \n",
       "\n",
       "     x1_std   x2_mean    x2_std  \n",
       "0  0.307068 -0.218063  0.045395  \n",
       "1  0.270489   0.22577  0.209123  \n",
       "2   0.00722  0.049739  0.215013  "
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_actors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "801943bf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Name</th>\n",
       "      <th>Timestep</th>\n",
       "      <th>x0</th>\n",
       "      <th>x1</th>\n",
       "      <th>x2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9276,1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9276</td>\n",
       "      <td>0.514415</td>\n",
       "      <td>-0.356119</td>\n",
       "      <td>0.147344</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9289,1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9289</td>\n",
       "      <td>-0.044279</td>\n",
       "      <td>0.030412</td>\n",
       "      <td>-0.322912</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9295,1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9295</td>\n",
       "      <td>0.227149</td>\n",
       "      <td>-0.138956</td>\n",
       "      <td>0.374034</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9305,1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9305</td>\n",
       "      <td>0.712255</td>\n",
       "      <td>-0.330702</td>\n",
       "      <td>0.5178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9316,1</th>\n",
       "      <td>APT1</td>\n",
       "      <td>9316</td>\n",
       "      <td>0.916359</td>\n",
       "      <td>-0.319975</td>\n",
       "      <td>0.294306</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99174,0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>99174</td>\n",
       "      <td>0.81886</td>\n",
       "      <td>-1.37145</td>\n",
       "      <td>-0.165163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99186,0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>99186</td>\n",
       "      <td>0.759486</td>\n",
       "      <td>-0.996092</td>\n",
       "      <td>-0.250518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99192,0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>99192</td>\n",
       "      <td>0.792642</td>\n",
       "      <td>-0.182136</td>\n",
       "      <td>-0.140075</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99194,0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>99194</td>\n",
       "      <td>0.130803</td>\n",
       "      <td>-0.193301</td>\n",
       "      <td>-0.193747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>99232,0</th>\n",
       "      <td>APT0</td>\n",
       "      <td>99232</td>\n",
       "      <td>0.49011</td>\n",
       "      <td>-0.418478</td>\n",
       "      <td>-0.281538</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>12418 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         Name Timestep        x0        x1        x2\n",
       "9276,1   APT1     9276  0.514415 -0.356119  0.147344\n",
       "9289,1   APT1     9289 -0.044279  0.030412 -0.322912\n",
       "9295,1   APT1     9295  0.227149 -0.138956  0.374034\n",
       "9305,1   APT1     9305  0.712255 -0.330702    0.5178\n",
       "9316,1   APT1     9316  0.916359 -0.319975  0.294306\n",
       "...       ...      ...       ...       ...       ...\n",
       "99174,0  APT0    99174   0.81886  -1.37145 -0.165163\n",
       "99186,0  APT0    99186  0.759486 -0.996092 -0.250518\n",
       "99192,0  APT0    99192  0.792642 -0.182136 -0.140075\n",
       "99194,0  APT0    99194  0.130803 -0.193301 -0.193747\n",
       "99232,0  APT0    99232   0.49011 -0.418478 -0.281538\n",
       "\n",
       "[12418 rows x 5 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_incidents"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5fa5eb8",
   "metadata": {},
   "source": [
    "# Describe dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a1fbb2fb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Name</th>\n",
       "      <th>Timestep</th>\n",
       "      <th>x0</th>\n",
       "      <th>x1</th>\n",
       "      <th>x2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>12418</td>\n",
       "      <td>12418</td>\n",
       "      <td>12418.00000</td>\n",
       "      <td>12418.000000</td>\n",
       "      <td>12418.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unique</th>\n",
       "      <td>3</td>\n",
       "      <td>11902</td>\n",
       "      <td>9432.00000</td>\n",
       "      <td>9467.000000</td>\n",
       "      <td>9397.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>top</th>\n",
       "      <td>APT1</td>\n",
       "      <td>66773</td>\n",
       "      <td>0.60363</td>\n",
       "      <td>-0.822319</td>\n",
       "      <td>-0.257797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freq</th>\n",
       "      <td>6585</td>\n",
       "      <td>3</td>\n",
       "      <td>6.00000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>6.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Name  Timestep           x0            x1            x2\n",
       "count   12418     12418  12418.00000  12418.000000  12418.000000\n",
       "unique      3     11902   9432.00000   9467.000000   9397.000000\n",
       "top      APT1     66773      0.60363     -0.822319     -0.257797\n",
       "freq     6585         3      6.00000      7.000000      6.000000"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_incidents.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "3084bf2c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Timestep</th>\n",
       "      <th>x0</th>\n",
       "      <th>x1</th>\n",
       "      <th>x2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>12418</td>\n",
       "      <td>12418.00000</td>\n",
       "      <td>12418.000000</td>\n",
       "      <td>12418.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unique</th>\n",
       "      <td>11902</td>\n",
       "      <td>9432.00000</td>\n",
       "      <td>9467.000000</td>\n",
       "      <td>9397.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>top</th>\n",
       "      <td>66773</td>\n",
       "      <td>0.60363</td>\n",
       "      <td>-0.822319</td>\n",
       "      <td>-0.257797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>freq</th>\n",
       "      <td>3</td>\n",
       "      <td>6.00000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>6.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Timestep           x0            x1            x2\n",
       "count      12418  12418.00000  12418.000000  12418.000000\n",
       "unique     11902   9432.00000   9467.000000   9397.000000\n",
       "top        66773      0.60363     -0.822319     -0.257797\n",
       "freq           3      6.00000      7.000000      6.000000"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_incidents[['Timestep'] + feature_names].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "98386874",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = matplotlib.pyplot.subplots(figsize=FIGSIZE)\n",
    "\n",
    "fg = seaborn.violinplot(data=df_incidents[feature_names], inner='box', color=seaborn.color_palette()[0])\n",
    "\n",
    "fig.savefig('feature-distributions.pdf', dpi=600, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "9553ad62",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
      "  with pd.option_context('mode.use_inf_as_na', True):\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "n_actors_shown = 4\n",
    "\n",
    "shown_actors = [f'APT{i}' for i in range(n_actors_shown)]\n",
    "\n",
    "fig, ax = matplotlib.pyplot.subplots(figsize=FIGSIZE)\n",
    "\n",
    "fg = seaborn.kdeplot(\n",
    "    data=df_incidents[df_incidents['Name'].apply(lambda x: x in shown_actors)],\n",
    "    x='Timestep',\n",
    "    hue=\"Name\",\n",
    "    hue_order=shown_actors,\n",
    ")\n",
    "\n",
    "fg.set_ylabel('Activity of threat actor')\n",
    "\n",
    "fig.savefig('threat-actor-activity.pdf', dpi=600, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "22895c75",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1498: FutureWarning: is_categorical_dtype is deprecated and will be removed in a future version. Use isinstance(dtype, CategoricalDtype) instead\n",
      "  if pd.api.types.is_categorical_dtype(vector):\n",
      "c:\\Users\\20173892\\AppData\\Local\\Programs\\Python\\venv\\Lib\\site-packages\\seaborn\\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
      "  with pd.option_context('mode.use_inf_as_na', True):\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x400 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = matplotlib.pyplot.subplots(figsize=FIGSIZE)\n",
    "\n",
    "fg = seaborn.histplot(\n",
    "    data=df_incidents['Name'].value_counts(), \n",
    "    kde=True,\n",
    ")\n",
    "\n",
    "fg.set_xlabel('Number of incidents associated with threat actor')\n",
    "fg.set_ylabel('Number of threat actors')\n",
    "\n",
    "fig.savefig('threat-actor-hist.pdf', dpi=600, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0877e81d",
   "metadata": {},
   "source": [
    "# Load/save dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "21f5c04a",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_incidents.to_csv('df_incidents_example.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "7380748f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_incidents = pandas.read_csv('df_incidents_example.csv')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "56d50a90-30c6-4b19-af42-98cedd98b2e8",
   "metadata": {},
   "source": [
    "# Split dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f97ea3e3-cee1-4fb2-8ee0-c84cae4bc1b0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4977, 7441)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_train = df_incidents[df_incidents['Timestep'] < EXPERIMENT_SIZE * TRAIN_SIZE]\n",
    "df_test = df_incidents[df_incidents['Timestep'] >= EXPERIMENT_SIZE * TRAIN_SIZE]\n",
    "\n",
    "len(df_train), len(df_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "a3514cf4-b411-427f-b3bb-5991caa8a8bd",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\20173892\\AppData\\Local\\Temp\\ipykernel_31012\\4255208019.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  df_train['Name'] = df_train['Name'].apply(lambda x: int(x[3:]))\n",
      "C:\\Users\\20173892\\AppData\\Local\\Temp\\ipykernel_31012\\4255208019.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  df_test['Name'] = df_test['Name'].apply(lambda x: int(x[3:]))\n"
     ]
    }
   ],
   "source": [
    "df_train['Name'] = df_train['Name'].apply(lambda x: int(x[3:]))\n",
    "df_test['Name'] = df_test['Name'].apply(lambda x: int(x[3:]))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "b1ea0dcb-e5b6-4143-89d9-730c7f1df894",
   "metadata": {},
   "source": [
    "## Run experiment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "9717da16-2f36-495a-bc0b-f86ece0389a7",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 3/3 [00:04<00:00,  1.38s/it]\n"
     ]
    }
   ],
   "source": [
    "models = [\n",
    "    copy.deepcopy(MODEL)\n",
    "    for _ in range(FEATURES)\n",
    "]\n",
    "\n",
    "for i in tqdm.tqdm(range(FEATURES)):\n",
    "    models[i].fit(df_train[[f'x{i}']].to_numpy(), df_train['Name'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "4732ffda-dff4-4ccc-aea7-7feab4df96d1",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 3/3 [00:00<00:00,  5.38it/s]\n"
     ]
    }
   ],
   "source": [
    "preds = []\n",
    "for i in tqdm.tqdm(range(FEATURES)):\n",
    "    preds.append(models[i].predict_proba(df_test[[f'x{i}']].to_numpy()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "7ba53b39-6497-4e19-91fe-d68514a2c9b1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using incident 689 as an example.\n",
      "According to the ground truth, this sample should be attributed to APT 2\n",
      "Prediction for feature 0 is given by: \n",
      "[0.01322612 0.46335697 0.52341691]\n",
      "Based on this prediction, the incident would be attributed to APT 2\n",
      "Prediction for feature 1 is given by: \n",
      "[0.02995507 0.82382383 0.14622111]\n",
      "Based on this prediction, the incident would be attributed to APT 1\n",
      "Prediction for feature 2 is given by: \n",
      "[0.48338059 0.02436817 0.49225124]\n",
      "Based on this prediction, the incident would be attributed to APT 2\n",
      "Combining predictions for features 0 and 1 with the logarithmic opinion pool results in: \n",
      "[0.00086381 0.83226883 0.16686736]\n",
      "Based on this prediction, the incident would be attributed to APT 1\n",
      "Combining predictions for features 0 and 2 with the logarithmic opinion pool results in: \n",
      "[0.02321972 0.04100851 0.93577177]\n",
      "Based on this prediction, the incident would be attributed to APT 2\n",
      "Combining predictions for features 1 and 2 with the logarithmic opinion pool results in: \n",
      "[0.13591839 0.18844123 0.67564038]\n",
      "Based on this prediction, the incident would be attributed to APT 2\n",
      "Combining intermediate results from the pairs with the linear opinion pool results in: \n",
      "[0.05333397 0.35390619 0.59275984]\n",
      "Based on this prediction, the incident would be attributed to APT 2\n"
     ]
    }
   ],
   "source": [
    "pool_name = 'Pairing Aggregator'\n",
    "\n",
    "incident_i = random.randint(0, len(df_test.index))\n",
    "print(f'Using incident {incident_i} as an example.')\n",
    "\n",
    "if not isinstance(df_test.iloc[incident_i]['Name'], str):\n",
    "    apt_i = df_test.iloc[incident_i]['Name']\n",
    "else:\n",
    "    apt_i = int(df_test.iloc[incident_i]['Name'][3:])\n",
    "\n",
    "print(f'According to the ground truth, this sample should be attributed to APT {apt_i}')\n",
    "\n",
    "for j in range(FEATURES):\n",
    "    print(f'Prediction for feature {j} is given by: ')\n",
    "    print(preds[j][i])\n",
    "    print(f'Based on this prediction, the incident would be attributed to APT {numpy.argmax(preds[j][i])}')\n",
    "\n",
    "X = numpy.array([preds[j][i] for j in range(FEATURES)])\n",
    "Y = []\n",
    "\n",
    "for i in range(X.shape[0]):\n",
    "    for j in range(i+1, X.shape[0]):\n",
    "        c = logarithmic_opinion_pool(numpy.array([X[i, :], X[j, :]]))\n",
    "        Y.append(c)\n",
    "        \n",
    "        print(f'Combining predictions for features {i} and {j} with the logarithmic opinion pool results in: ')\n",
    "        print(c)\n",
    "        print(f'Based on this prediction, the incident would be attributed to APT {numpy.argmax(c)}')\n",
    "\n",
    "Y = numpy.array(Y)\n",
    "\n",
    "c = linear_opinion_pool(Y)\n",
    "        \n",
    "print(f'Combining intermediate results from the pairs with the linear opinion pool results in: ')\n",
    "print(c)\n",
    "print(f'Based on this prediction, the incident would be attributed to APT {numpy.argmax(c)}')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "language": "python",
   "name": "venv"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}