sec5.3-service_availability.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from util.plotting import *\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from datetime import date, datetime, time, timedelta\n",
    "import json\n",
    "import seaborn as sns\n",
    "from matplotlib.colors import ListedColormap, BoundaryNorm\n",
    "pd.set_option('display.max_columns', None)\n",
    "plt.rcParams['text.usetex'] = True\n",
    "plt.rcParams['font.family'] = 'libertine'\n",
    "plt.rcParams['font.weight'] = 'bold'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load outage data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>Date</th>\n",
       "      <th>Outage_Color</th>\n",
       "      <th>Incidents</th>\n",
       "      <th>Service</th>\n",
       "      <th>incident_count</th>\n",
       "      <th>partial_outage_flag</th>\n",
       "      <th>partial_outage_minutes</th>\n",
       "      <th>major_outage_flag</th>\n",
       "      <th>major_outage_minutes</th>\n",
       "      <th>outage_flag</th>\n",
       "      <th>total_outage_minutes</th>\n",
       "      <th>scaled_total_outage_minutes</th>\n",
       "      <th>outage_impact_level</th>\n",
       "      <th>provider</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2024-06-01</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2024-06-02</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2024-06-03</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2024-06-04</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2024-06-05</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date Outage_Color Incidents     Service  incident_count  \\\n",
       "0 2024-06-01      #10a37f        []  API-OpenAI               0   \n",
       "1 2024-06-02      #10a37f        []  API-OpenAI               0   \n",
       "2 2024-06-03      #10a37f        []  API-OpenAI               0   \n",
       "3 2024-06-04      #10a37f        []  API-OpenAI               0   \n",
       "4 2024-06-05      #10a37f        []  API-OpenAI               0   \n",
       "\n",
       "   partial_outage_flag  partial_outage_minutes  major_outage_flag  \\\n",
       "0                    0                       0                  0   \n",
       "1                    0                       0                  0   \n",
       "2                    0                       0                  0   \n",
       "3                    0                       0                  0   \n",
       "4                    0                       0                  0   \n",
       "\n",
       "   major_outage_minutes  outage_flag  total_outage_minutes  \\\n",
       "0                     0            0                     0   \n",
       "1                     0            0                     0   \n",
       "2                     0            0                     0   \n",
       "3                     0            0                     0   \n",
       "4                     0            0                     0   \n",
       "\n",
       "   scaled_total_outage_minutes  outage_impact_level provider  \n",
       "0                          0.0                    0   openai  \n",
       "1                          0.0                    0   openai  \n",
       "2                          0.0                    0   openai  \n",
       "3                          0.0                    0   openai  \n",
       "4                          0.0                    0   openai  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>Date</th>\n",
       "      <th>Outage_Color</th>\n",
       "      <th>Incidents</th>\n",
       "      <th>Service</th>\n",
       "      <th>incident_count</th>\n",
       "      <th>partial_outage_flag</th>\n",
       "      <th>partial_outage_minutes</th>\n",
       "      <th>major_outage_flag</th>\n",
       "      <th>major_outage_minutes</th>\n",
       "      <th>outage_flag</th>\n",
       "      <th>total_outage_minutes</th>\n",
       "      <th>scaled_total_outage_minutes</th>\n",
       "      <th>outage_impact_level</th>\n",
       "      <th>provider</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2024-06-01</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>DALL·E</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2024-06-02</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>DALL·E</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2024-06-03</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>DALL·E</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2024-06-04</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>DALL·E</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2024-06-05</td>\n",
       "      <td>#10a37f</td>\n",
       "      <td>[]</td>\n",
       "      <td>DALL·E</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>openai</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date Outage_Color Incidents Service  incident_count  \\\n",
       "0 2024-06-01      #10a37f        []  DALL·E               0   \n",
       "1 2024-06-02      #10a37f        []  DALL·E               0   \n",
       "2 2024-06-03      #10a37f        []  DALL·E               0   \n",
       "3 2024-06-04      #10a37f        []  DALL·E               0   \n",
       "4 2024-06-05      #10a37f        []  DALL·E               0   \n",
       "\n",
       "   partial_outage_flag  partial_outage_minutes  major_outage_flag  \\\n",
       "0                    0                       0                  0   \n",
       "1                    0                       0                  0   \n",
       "2                    0                       0                  0   \n",
       "3                    0                       0                  0   \n",
       "4                    0                       0                  0   \n",
       "\n",
       "   major_outage_minutes  outage_flag  total_outage_minutes  \\\n",
       "0                     0            0                     0   \n",
       "1                     0            0                     0   \n",
       "2                     0            0                     0   \n",
       "3                     0            0                     0   \n",
       "4                     0            0                     0   \n",
       "\n",
       "   scaled_total_outage_minutes  outage_impact_level provider  \n",
       "0                          0.0                    0   openai  \n",
       "1                          0.0                    0   openai  \n",
       "2                          0.0                    0   openai  \n",
       "3                          0.0                    0   openai  \n",
       "4                          0.0                    0   openai  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Load data\n",
    "path = f'data/clean/outage/2024-08-31/outage_unexploded.csv'\n",
    "df = pd.read_csv(path, parse_dates=['Date'])\n",
    "# rename Service\n",
    "for service in df['Service'].unique():\n",
    "    if service == \"api\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"API-OpenAI\"\n",
    "    elif service == \"chatgpt\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"ChatGPT\"\n",
    "    elif service == \"labs\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"DALL·E\"\n",
    "    elif service == \"playground\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"Playground\"\n",
    "    elif service == \"api.anthropic.com\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"API-Anthropic\"\n",
    "    elif service == \"claude.ai\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"Claude\"\n",
    "    elif service == \"console.anthropic.com\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"Console\"\n",
    "    elif service == \"character.ai\":\n",
    "        df.loc[df['Service'] == service, 'Service'] = \"Character.AI\"\n",
    "\n",
    "service_lst = ['API-OpenAI', 'ChatGPT', 'DALL·E', 'Playground', 'API-Anthropic', 'Claude', 'Console', 'Character.AI']\n",
    "dfs = {}\n",
    "# filter by service\n",
    "for service in service_lst:\n",
    "    dfs[service] = df[df['Service'] == service]\n",
    "    dfs[service].reset_index(drop=True, inplace=True)\n",
    "display(dfs['API-OpenAI'].head())\n",
    "\n",
    "provider_lst = ['openai', 'anthropic', 'characterai']\n",
    "dfs_provider = {}\n",
    "# filter by provider\n",
    "for provider in provider_lst:\n",
    "    dfs_provider[provider] = df[df['provider'] == provider]\n",
    "    dfs_provider[provider].reset_index(drop=True, inplace=True)\n",
    "display(dfs_provider['openai'].head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Min daily availability:\n",
      "API-OpenAI: 70.35%\n",
      "ChatGPT: 64.31%\n",
      "DALL·E: 54.17%\n",
      "Playground: 80.69%\n",
      "API-Anthropic: 51.46%\n",
      "Claude: 41.87%\n",
      "Console: 41.87%\n",
      "Character.AI: 59.72%\n",
      "\n",
      "Min scaled daily availability:\n",
      "API-OpenAI: 80.57%\n",
      "ChatGPT: 77.15%\n",
      "DALL·E: 74.17%\n",
      "Playground: 82.57%\n",
      "API-Anthropic: 85.44%\n",
      "Claude: 82.02%\n",
      "Console: 82.56%\n",
      "Character.AI: 85.33%\n"
     ]
    }
   ],
   "source": [
    "# calculate \"daily_availability\" column\n",
    "for service in service_lst:\n",
    "    df = dfs[service]\n",
    "    df['daily_availability'] = 1 - df['total_outage_minutes'] / (24 * 60)\n",
    "    df['scaled_daily_availability'] = 1 - df['scaled_total_outage_minutes'] / (24 * 60)\n",
    "    df['daily_availability'] = df['daily_availability'] * 100\n",
    "    df['scaled_daily_availability'] = df['scaled_daily_availability'] * 100\n",
    "\n",
    "dfs['API-OpenAI'].head()\n",
    "\n",
    "# display the min daily availability in 2 decimal places\n",
    "print(\"Min daily availability:\")\n",
    "for service in service_lst:\n",
    "    df = dfs[service]\n",
    "    min_availability = df['daily_availability'].min()\n",
    "    print(f\"{service}: {min_availability:.2f}%\")\n",
    "\n",
    "# display the min scaled daily availability in 2 decimal places\n",
    "print()\n",
    "print(\"Min scaled daily availability:\")\n",
    "for service in service_lst:\n",
    "    df = dfs[service]\n",
    "    min_availability = df['scaled_daily_availability'].min()\n",
    "    print(f\"{service}: {min_availability:.2f}%\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## fig. daily availability by scaled total outage minutes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>Date</th>\n",
       "      <th>API-OpenAI</th>\n",
       "      <th>ChatGPT</th>\n",
       "      <th>DALL·E</th>\n",
       "      <th>Playground</th>\n",
       "      <th>API-Anthropic</th>\n",
       "      <th>Claude</th>\n",
       "      <th>Console</th>\n",
       "      <th>Character.AI</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2021-02-11</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2021-02-12</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2021-02-13</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2021-02-14</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2021-02-15</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1293</th>\n",
       "      <td>2024-08-27</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1294</th>\n",
       "      <td>2024-08-28</td>\n",
       "      <td>97.083333</td>\n",
       "      <td>96.583333</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1295</th>\n",
       "      <td>2024-08-29</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1296</th>\n",
       "      <td>2024-08-30</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1297</th>\n",
       "      <td>2024-08-31</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>100.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1298 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           Date  API-OpenAI     ChatGPT  DALL·E  Playground  API-Anthropic  \\\n",
       "0    2021-02-11  100.000000         NaN     NaN         NaN            NaN   \n",
       "1    2021-02-12  100.000000         NaN     NaN         NaN            NaN   \n",
       "2    2021-02-13  100.000000         NaN     NaN         NaN            NaN   \n",
       "3    2021-02-14  100.000000         NaN     NaN         NaN            NaN   \n",
       "4    2021-02-15  100.000000         NaN     NaN         NaN            NaN   \n",
       "...         ...         ...         ...     ...         ...            ...   \n",
       "1293 2024-08-27  100.000000  100.000000   100.0       100.0          100.0   \n",
       "1294 2024-08-28   97.083333   96.583333   100.0       100.0          100.0   \n",
       "1295 2024-08-29  100.000000  100.000000   100.0       100.0          100.0   \n",
       "1296 2024-08-30  100.000000  100.000000   100.0       100.0          100.0   \n",
       "1297 2024-08-31  100.000000  100.000000   100.0       100.0          100.0   \n",
       "\n",
       "      Claude  Console  Character.AI  \n",
       "0        NaN      NaN           NaN  \n",
       "1        NaN      NaN           NaN  \n",
       "2        NaN      NaN           NaN  \n",
       "3        NaN      NaN           NaN  \n",
       "4        NaN      NaN           NaN  \n",
       "...      ...      ...           ...  \n",
       "1293   100.0    100.0         100.0  \n",
       "1294   100.0    100.0         100.0  \n",
       "1295   100.0    100.0         100.0  \n",
       "1296   100.0    100.0         100.0  \n",
       "1297   100.0    100.0         100.0  \n",
       "\n",
       "[1298 rows x 9 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# preapre data for heatmap\n",
    "# creata a dataframe with date from 2021-02-11 to 2024-05-31\n",
    "start_date = date(2021, 2, 11)\n",
    "# start_date = date(2023, 2, 1)\n",
    "end_date = date(2024, 8, 31)\n",
    "delta = timedelta(days=1)\n",
    "dates = []\n",
    "while start_date <= end_date:\n",
    "    dates.append(start_date)\n",
    "    start_date += delta\n",
    "\n",
    "dates = pd.DataFrame(dates, columns=['Date'])\n",
    "dates['Date'] = pd.to_datetime(dates['Date'])\n",
    "dates\n",
    "\n",
    "# dfs[service] left join dates on 'Date' with only 'daily_availability' column\n",
    "for service in service_lst:\n",
    "    df = dfs[service]\n",
    "    df = pd.merge(dates, df, on='Date', how='left')\n",
    "    # select only 'Date' and 'daily_availability' columns\n",
    "    df = df[['Date', 'scaled_daily_availability']]\n",
    "    dfs[service] = df\n",
    "\n",
    "# prepare data for plotting\n",
    "data = {\n",
    "    'Date': dfs['API-OpenAI']['Date'],\n",
    "    'API-OpenAI': dfs['API-OpenAI']['scaled_daily_availability'],\n",
    "    'ChatGPT': dfs['ChatGPT']['scaled_daily_availability'],\n",
    "    'DALL·E': dfs['DALL·E']['scaled_daily_availability'],\n",
    "    'Playground': dfs['Playground']['scaled_daily_availability'],\n",
    "    'API-Anthropic': dfs['API-Anthropic']['scaled_daily_availability'],\n",
    "    'Claude': dfs['Claude']['scaled_daily_availability'],\n",
    "    'Console': dfs['Console']['scaled_daily_availability'],\n",
    "    'Character.AI': dfs['Character.AI']['scaled_daily_availability']\n",
    "}\n",
    "\n",
    "df = pd.DataFrame(data)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1600x300 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# fig. daily availability by downtime\n",
    "set_one_column_figure_font(height_multiplier=1.0)\n",
    "\n",
    "# Melt the DataFrame to have services in one column\n",
    "df_melted = df.melt(id_vars='Date', var_name='Service', value_name='Availability', ignore_index=True)\n",
    "\n",
    "custom_colors = ['#ef4146', '#ea7711', '#f3ac36', '#c0aa47', '#10a37f']\n",
    "cmap = ListedColormap(custom_colors)\n",
    "custom_bins = [50, 90, 99, 99.9, 99.99, 100] \n",
    "norm = BoundaryNorm(custom_bins, cmap.N)\n",
    "\n",
    "# Create the heatmap\n",
    "plt.figure(figsize=(16, 3))\n",
    "heatmap_data = df_melted.pivot(index=\"Service\", columns=\"Date\", values=\"Availability\")\n",
    "heatmap_data = heatmap_data.reindex(['API-OpenAI', 'ChatGPT', 'DALL·E', 'Playground', 'API-Anthropic', 'Claude', 'Console', 'Character.AI']) \n",
    "mask = heatmap_data.isnull()\n",
    "\n",
    "sns.heatmap(heatmap_data, cmap=cmap, norm=norm, annot=False, fmt=\".0f\", cbar_kws={'ticks': custom_bins})\n",
    "\n",
    "# Set x-ticks the start of each month\n",
    "date_range = pd.date_range(start=df['Date'].min(), end=df['Date'].max(), freq='MS')\n",
    "date_labels = [date.strftime('%Y-%m') for date in date_range]\n",
    "plt.xticks([heatmap_data.columns.get_loc(date) for date in date_range])\n",
    "plt.gca().set_xticklabels(date_labels, rotation=30)\n",
    "\n",
    "plt.savefig('plot/sec5-temporal_analysis/fig-daily_availability_by_scaled_outage_minutes.pdf', bbox_inches='tight')\n",
    "plt.savefig('plot/sec5-temporal_analysis/fig-daily_availability_by_scaled_outage_minutes.png', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## table. daily availability by scaled outage minutes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>Service</th>\n",
       "      <th>Min</th>\n",
       "      <th>Max</th>\n",
       "      <th>Mean</th>\n",
       "      <th>Median</th>\n",
       "      <th>=100%</th>\n",
       "      <th>&gt;=99.999%</th>\n",
       "      <th>&gt;=99.99%</th>\n",
       "      <th>&gt;=99.9%</th>\n",
       "      <th>&gt;=99%</th>\n",
       "      <th>&gt;=90%</th>\n",
       "      <th>&lt;90%</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>API-OpenAI</td>\n",
       "      <td>80.57%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.82%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>92.68%</td>\n",
       "      <td>92.68%</td>\n",
       "      <td>92.68%</td>\n",
       "      <td>92.84%</td>\n",
       "      <td>95.3%</td>\n",
       "      <td>99.77%</td>\n",
       "      <td>0.23%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ChatGPT</td>\n",
       "      <td>77.15%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.66%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>88.85%</td>\n",
       "      <td>88.85%</td>\n",
       "      <td>88.85%</td>\n",
       "      <td>89.2%</td>\n",
       "      <td>93.1%</td>\n",
       "      <td>99.12%</td>\n",
       "      <td>0.88%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>DALL·E</td>\n",
       "      <td>74.17%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.78%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>95.88%</td>\n",
       "      <td>95.88%</td>\n",
       "      <td>95.88%</td>\n",
       "      <td>95.88%</td>\n",
       "      <td>96.77%</td>\n",
       "      <td>99.28%</td>\n",
       "      <td>0.72%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Playground</td>\n",
       "      <td>82.57%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.94%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>98.08%</td>\n",
       "      <td>98.08%</td>\n",
       "      <td>98.08%</td>\n",
       "      <td>98.16%</td>\n",
       "      <td>98.88%</td>\n",
       "      <td>99.84%</td>\n",
       "      <td>0.16%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>API-Anthropic</td>\n",
       "      <td>85.44%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.92%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>94.02%</td>\n",
       "      <td>94.02%</td>\n",
       "      <td>94.02%</td>\n",
       "      <td>94.26%</td>\n",
       "      <td>97.61%</td>\n",
       "      <td>99.76%</td>\n",
       "      <td>0.24%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Claude</td>\n",
       "      <td>82.02%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.84%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>93.06%</td>\n",
       "      <td>93.06%</td>\n",
       "      <td>93.06%</td>\n",
       "      <td>93.06%</td>\n",
       "      <td>97.13%</td>\n",
       "      <td>99.52%</td>\n",
       "      <td>0.48%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Console</td>\n",
       "      <td>82.56%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.89%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>93.54%</td>\n",
       "      <td>93.54%</td>\n",
       "      <td>93.54%</td>\n",
       "      <td>93.54%</td>\n",
       "      <td>97.61%</td>\n",
       "      <td>99.76%</td>\n",
       "      <td>0.24%</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Character.AI</td>\n",
       "      <td>85.33%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>99.59%</td>\n",
       "      <td>100.0%</td>\n",
       "      <td>90.88%</td>\n",
       "      <td>90.88%</td>\n",
       "      <td>90.88%</td>\n",
       "      <td>91.19%</td>\n",
       "      <td>94.03%</td>\n",
       "      <td>98.11%</td>\n",
       "      <td>1.89%</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Service     Min     Max    Mean  Median   =100% >=99.999% >=99.99%  \\\n",
       "0     API-OpenAI  80.57%  100.0%  99.82%  100.0%  92.68%    92.68%   92.68%   \n",
       "1        ChatGPT  77.15%  100.0%  99.66%  100.0%  88.85%    88.85%   88.85%   \n",
       "2         DALL·E  74.17%  100.0%  99.78%  100.0%  95.88%    95.88%   95.88%   \n",
       "3     Playground  82.57%  100.0%  99.94%  100.0%  98.08%    98.08%   98.08%   \n",
       "4  API-Anthropic  85.44%  100.0%  99.92%  100.0%  94.02%    94.02%   94.02%   \n",
       "5         Claude  82.02%  100.0%  99.84%  100.0%  93.06%    93.06%   93.06%   \n",
       "6        Console  82.56%  100.0%  99.89%  100.0%  93.54%    93.54%   93.54%   \n",
       "7   Character.AI  85.33%  100.0%  99.59%  100.0%  90.88%    90.88%   90.88%   \n",
       "\n",
       "  >=99.9%   >=99%   >=90%   <90%  \n",
       "0  92.84%   95.3%  99.77%  0.23%  \n",
       "1   89.2%   93.1%  99.12%  0.88%  \n",
       "2  95.88%  96.77%  99.28%  0.72%  \n",
       "3  98.16%  98.88%  99.84%  0.16%  \n",
       "4  94.26%  97.61%  99.76%  0.24%  \n",
       "5  93.06%  97.13%  99.52%  0.48%  \n",
       "6  93.54%  97.61%  99.76%  0.24%  \n",
       "7  91.19%  94.03%  98.11%  1.89%  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# table. service availability\n",
    "# create a dataframe with one column for each service\n",
    "ava_100 = []\n",
    "ava_99999 = []\n",
    "ava_9999 = []\n",
    "ava_999 = []\n",
    "ava_99 = []\n",
    "ava_90 = []\n",
    "ava_50 = []\n",
    "\n",
    "for service in service_lst:\n",
    "    df = dfs[service]\n",
    "    ava_100.append(df[df['scaled_daily_availability'] == 100].shape[0] /  df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "    ava_99999.append(df[df['scaled_daily_availability'] >= 99.999].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)  \n",
    "    ava_9999.append(df[df['scaled_daily_availability'] >= 99.99].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "    ava_999.append(df[df['scaled_daily_availability'] >= 99.9].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "    ava_99.append(df[df['scaled_daily_availability'] >= 99].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "    ava_90.append(df[df['scaled_daily_availability'] >= 90].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "    ava_50.append(df[df['scaled_daily_availability'] < 90].shape[0] / df['scaled_daily_availability'].dropna().shape[0] * 100)\n",
    "\n",
    "data = {\n",
    "    'Service': ['API-OpenAI', 'ChatGPT', 'DALL·E', 'Playground', 'API-Anthropic', 'Claude', 'Console', 'Character.AI'],\n",
    "    'Min': [dfs['API-OpenAI']['scaled_daily_availability'].min(), dfs['ChatGPT']['scaled_daily_availability'].min(), dfs['DALL·E']['scaled_daily_availability'].min(), dfs['Playground']['scaled_daily_availability'].min(), dfs['API-Anthropic']['scaled_daily_availability'].min(), dfs['Claude']['scaled_daily_availability'].min(), dfs['Console']['scaled_daily_availability'].min(), dfs['Character.AI']['scaled_daily_availability'].min()],\n",
    "    'Max': [dfs['API-OpenAI']['scaled_daily_availability'].max(), dfs['ChatGPT']['scaled_daily_availability'].max(), dfs['DALL·E']['scaled_daily_availability'].max(), dfs['Playground']['scaled_daily_availability'].max(), dfs['API-Anthropic']['scaled_daily_availability'].max(), dfs['Claude']['scaled_daily_availability'].max(), dfs['Console']['scaled_daily_availability'].max(), dfs['Character.AI']['scaled_daily_availability'].max()],\n",
    "    'Mean': [dfs['API-OpenAI']['scaled_daily_availability'].mean(), dfs['ChatGPT']['scaled_daily_availability'].mean(), dfs['DALL·E']['scaled_daily_availability'].mean(), dfs['Playground']['scaled_daily_availability'].mean(), dfs['API-Anthropic']['scaled_daily_availability'].mean(), dfs['Claude']['scaled_daily_availability'].mean(), dfs['Console']['scaled_daily_availability'].mean(), dfs['Character.AI']['scaled_daily_availability'].mean()],\n",
    "    'Median': [dfs['API-OpenAI']['scaled_daily_availability'].median(), dfs['ChatGPT']['scaled_daily_availability'].median(), dfs['DALL·E']['scaled_daily_availability'].median(), dfs['Playground']['scaled_daily_availability'].median(), dfs['API-Anthropic']['scaled_daily_availability'].median(), dfs['Claude']['scaled_daily_availability'].median(), dfs['Console']['scaled_daily_availability'].median(), dfs['Character.AI']['scaled_daily_availability'].median()],\n",
    "    '=100%': ava_100,\n",
    "    '>=99.999%': ava_99999,\n",
    "    '>=99.99%': ava_9999,\n",
    "    '>=99.9%': ava_999,\n",
    "    '>=99%': ava_99,\n",
    "    '>=90%': ava_90,\n",
    "    '<90%': ava_50\n",
    "}\n",
    "\n",
    "df_table_availability = pd.DataFrame(data)\n",
    "\n",
    "df_table_availability = df_table_availability.round(2)\n",
    "for column in df_table_availability.columns:\n",
    "    if column != 'Service':\n",
    "        df_table_availability[column] = df_table_availability[column].astype(str) + '%'\n",
    "\n",
    "df_table_availability"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\\begin{tabular}{llllllllllll}\n",
      "\\toprule\n",
      "Service & Min & Max & Mean & Median & =100% & >=99.999% & >=99.99% & >=99.9% & >=99% & >=90% & <90% \\\\\n",
      "\\midrule\n",
      "API-OpenAI & 80.57% & 100.0% & 99.82% & 100.0% & 92.68% & 92.68% & 92.68% & 92.84% & 95.3% & 99.77% & 0.23% \\\\\n",
      "ChatGPT & 77.15% & 100.0% & 99.66% & 100.0% & 88.85% & 88.85% & 88.85% & 89.2% & 93.1% & 99.12% & 0.88% \\\\\n",
      "DALL·E & 74.17% & 100.0% & 99.78% & 100.0% & 95.88% & 95.88% & 95.88% & 95.88% & 96.77% & 99.28% & 0.72% \\\\\n",
      "Playground & 82.57% & 100.0% & 99.94% & 100.0% & 98.08% & 98.08% & 98.08% & 98.16% & 98.88% & 99.84% & 0.16% \\\\\n",
      "API-Anthropic & 85.44% & 100.0% & 99.92% & 100.0% & 94.02% & 94.02% & 94.02% & 94.26% & 97.61% & 99.76% & 0.24% \\\\\n",
      "Claude & 82.02% & 100.0% & 99.84% & 100.0% & 93.06% & 93.06% & 93.06% & 93.06% & 97.13% & 99.52% & 0.48% \\\\\n",
      "Console & 82.56% & 100.0% & 99.89% & 100.0% & 93.54% & 93.54% & 93.54% & 93.54% & 97.61% & 99.76% & 0.24% \\\\\n",
      "Character.AI & 85.33% & 100.0% & 99.59% & 100.0% & 90.88% & 90.88% & 90.88% & 91.19% & 94.03% & 98.11% & 1.89% \\\\\n",
      "\\bottomrule\n",
      "\\end{tabular}\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# save the table\n",
    "df_table_availability.to_csv('plot/backup_table_csv/table-service_availability.csv', index=False)\n",
    "# to latex\n",
    "print(df_table_availability.to_latex(index=False)) # copy and let chatgpt add \\ before %"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.12.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}