{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Итоговый проект: отчёт\n",
    "\n",
    "Отчёт о годе «Кофейной лавки» по собранной таблице: динамика по месяцам, прибыль и возвраты по категориям, города и средний чек, сводная «категории × города», сегменты и лучшие покупатели, графики и выводы.\n",
    "\n",
    "Курс pandas · урок `pd-final-report` · [урок на сайте](https://ds-edu-trainer.vercel.app/courses/pandas/final-report).\n",
    "\n",
    "Выполняйте ячейки по порядку (Shift+Enter), начиная с первой — она готовит урок. После каждого упражнения идёт\n",
    "ячейка проверки: она печатает ✓ / ✗ и итог."
   ],
   "id": "intro"
  },
  {
   "cell_type": "code",
   "metadata": {
    "cellView": "form",
    "jupyter": {
     "source_hidden": true
    }
   },
   "source": [
    "# @title Подготовка урока — выполните эту ячейку первой\n",
    "# Проверяет версии пакетов, записывает файлы данных в папку data/ и определяет _check — проверку упражнений.\n",
    "import base64, os, zlib\n",
    "\n",
    "for _name, _min in {\"numpy\":\"2.0\",\"pandas\":\"3.0\"}.items():\n",
    "    try:\n",
    "        _have = __import__(_name).__version__\n",
    "    except ImportError:\n",
    "        print(f\"Нет пакета {_name}: выполните %pip install {_name} и перезапустите ядро\")\n",
    "        continue\n",
    "    if tuple(int(x) for x in _have.split(\".\")[:2]) < tuple(int(x) for x in _min.split(\".\")):\n",
    "        print(f\"Урок написан для {_name} {_min} и новее, у вас {_have}: выполните %pip install -U {_name} и перезапустите ядро\")\n",
    "\n",
    "_DATA = {\n",
    "    \"orders.csv\": \"\"\"\n",
    "eNqVfUuOdjFu3TxrKQMiqee4N9A7CAx3AzGQtBOne+CZ7RV50oEdJ/EW/t5R8En3QfKeS1XNCoVzJF191IuHov7u73/3+7//z3/7\n",
    "u6/f/fUff//1N3/6n3/8u/+2/vHf//7vfvenv/nj58//8ae//sMf//aP//D1N//lr//wh9//1/9EKSX64sTlrxL9VaKv36RRvn5L\n",
    "/Yu/fv2fX//yl3/89b9//fkv//zrP379+68///q3v/zTGyWNL4oobCllfP2W0ofyH3/5x1//+uvff/2/X//r159//d9f//rrz2+E\n",
    "VCOCGAIn+fptki/6+ss//fqXX//2l39+BX0+1oDyDeKv36T86RH59Mhb1cUSRPZdWA2FJM0urLYhzYJa/fpt4i95b0i/CfL1m88P\n",
    "9VsKCcMS2ueXbF/1lUDJEnr6+i3VTw2vn0pkKYNnd4YUthT+/Ewl+PEhgYKfDBJy0FOkLCd//YZozK6NfmPKhvL5zeOfj4qto6VJ\n",
    "COtQdlTO72jhUKRmKdLnp9uBQt2C+hpNbQ+qX9mClInVr98wtznkop+fk6GksX7+918TEmrQ05AwAgNjsoTazznh/SsAJclXjijs\n",
    "mvUZvxR9ByAkir5DLKHlg6B/Mgiqbpbk7H5XAmYEQX6+5WJAxPlcG947Spl9+5S7HyncDGX9HC3qKkQIjQoRcmS2avz0r9/QSAfB\n",
    "9A4CNTfIeBhQmrNT9T8GAM1hpUGSbHU1T5DsQb46IQBqvjq2bZL1q0SzA6Js1hNRBj0+lHTMxKYp2YFkO8NDSoqbgig5rqUYCpW1\n",
    "3XKtr7bcvHY2tAc9frdmq6ttTkLRcESUzYZQbrOm9Fk76mEdP6TUuJZhKCxts5HIyRE+c1FzPYRA83MNiGxrZzeyM38EevxqmQ2I\n",
    "qG/XeEgpYe9msZQ5z0Qb7pxfeuFHhBrMj7nYJs0zgPguRCC/DuXb8Omz7+e6XS8gJba23Axl9vrTeADo+VHdggpNUNmDyK0MCPQ0\n",
    "w2FBmcEaA0HdTUMlGVBq/TiFvfcaolANf5tC7rdp23kXUmILKGwpUua3vG8AitgvqQ0ssAj0OSJZkBoqfOzAPntWCyoW1AfYpiEQ\n",
    "JffTlmpANCowEgR6GG5pDpTOaf2tz5R9yodQwH4DgvzZvgwLSm0zvdRkCXXvxqhkKJ8f87P3yO91AMLcAP+IEB0aK9uv+OyekzhP\n",
    "AgSNb4A4uV6uynw/h9reNtvomi2hpGO7ZkotFvQ5xj7sHYES+/ZVA6JjX+FAzZbUyjG9vn5EN4R1iA1tSxlj+dhWPc40phkI5PcR\n",
    "LRnQx53y3Gw0ZTWfI1RNm7NKY0uYq5CfzCHITwtNDGhNehwYRMuW0PJ22CEKxTugViylp42dtmoI6yf283Jrtkf4XLyDhgDK5rjS\n",
    "uuv3MXuIf0qJN+1NWV8798Z+pPRkQCQ8jytR6ztZCu32hYiwluY3AhvCmoMkqgERIjdnF/fZyEcGQX7z3rOrms7z3lvVgBB/XLHN\n",
    "GLw9HvRqKdOxXH5MWa6pd4qy/H7O1z36km4I5663RnUMQ2FCToaRbENK3rogB1lKRmduCPJL21C29zkbM21HJqTEe9UhhvKRJNYC\n",
    "YJqSLSjT1omJKLuuK7b1H2fLRzZ4/dlHtXWUdjTLtB2AHmvRaA6UN2MHEebu+JVwmyinczWO9nOQUH5IoOjUMYYhUM5g4YIgN6FR\n",
    "SgZ0u3RfqoYECnb7pJTD+XGlHxOPaQYCNQ9iVzU/j1Gk5D6eQmU/f6036yUl/jGdfTB85cWASNaEUi2o2pKkPV3tGFT8Z9xWynx6\n",
    "PCn+DERJnx3uO6VbytwU8/siCQk0ol9/GMLapXbfIwDkZwNSMuQHxOnz6/tuU8LjLCmP55yIQexBbKtj2v4ASi+c5RI4nJNSCFnO\n",
    "vWEwVZJSCD+EZS71p4RgNsGEEtVQ7TccrsZsP1SZ46Vs5mgRxJR47Co9clJoI/GQ0iZ5ngPHTn3AlHDrREps5HyKVcGagQnB8ZSU\n",
    "1DgJcwftdisQ5FdQDMoepIbD59zaOrBtBHpM0LCk4UFiQYwmUgh6tCkb0FTkoygUUtIk13NHyl/tRwSi6LeuhrAkmfJTAkXf0Cxh\n",
    "tw0npTjy5wRC7SnakhITPyBO7fDbvY8DpRqucsfuoEJKQ+R2zqJ+nkcg8lO4Ug15HhT6phcgoQUzpxIZPwRiQW1FIPFtza7qejj2\n",
    "X6suttS+YrjoR4TIy0hKOZyEVo9Qj58QKFqpYA3RXlKJjtzPo0m0W1X6IU8dte/iDCBl9dM75R4Y8tlpc3meSEkpiHLqkhSKS5CS\n",
    "QpkEU0Zcyz2W5JIwIquHhB4R2BLGxlVJSpgUOlxQ4Z7n0hl5Bvl9fpG1qzK/AAB5NYMubXGBEq81LEcdWG25NT+d1HRpgGe5shO/\n",
    "IYUk3NDDWnr883dLkR6HPtIlCh6Ew/n0TrjUPl4xjUyxs48u4Y6PaMy6HYKQEo+uS+s7KLKfGyAl9NJAymYbfMmFfAejpu621yXb\n",
    "Tv1sSknicgEl9mRTUaNBbs+k/JRCccOqoXy2iGFkJpVmCZUPD6vpIQB6+D9Kt639TN2xzAApm1m5DNeUDGaemly5a1y1LYj8xriS\n",
    "rS7LUwCmqgwzH7u5Z0kA5H2iVJW9llNvDuQgSAg3ZpAQbYBrNgSG5wcI8k6nWgzo4xd4hNlRVQb8EZJS33lvMSUeWbVZynHecU3p\n",
    "FiQZffkwoG1gJCREHkpqyX1eO+eanxAiz1JTRt50GLT+UAjyXdbYgmoGJ1II8kfpJhY0+nZqgJR4AmrZNUX2lGIpnbZHt1YtZfr0\n",
    "o1UbEdIIzgutGcLyBMQrBaTEHqKmBkQ/PCsPRwkCPdaKNgxohaTEo7wnW26SY/5865NOhrBkSxcbQ51dqW0XUEpdbNulABdrz7Zy\n",
    "zodmakDFgQRstXu1oBmk7v2wENS/UxLbsDjqyiwuycoPTgjyp7TeLei6hmJAtyHccZPRAQgSoi3zSIawJmh/pIAg71gYt0HNED6u\n",
    "wCc32III/aZDLKgmsBIi0GPCHdmAPpdgdnviUQxlOil9oBKN21boPGhE8aw0bpMgORfL6Icc3RLmPOFX1zEMaM3P3wK5X5iT+oWn\n",
    "d7s8N3IQNM/vBkQGtBZwt8ZxYlvS9Mdk3yZ5lhSvcZzU7/3ZQx3bo2LLRaDxHZB4UDGg8+Ihb0F+8HCqFtT4lLvevxVQUv24j98p\n",
    "zTVFdmdXTBnRes6p266bV27I/7jDlZsOke29XFLmV4+QmFgoYyJHabuTN6SkGvlCMCUMLWFiQ1k7bD9KSGy518XDLcjtTpmyAaXp\n",
    "5hoeVCyI5Dhd7UB+bWOqFjTazmXA1OxnXDeFTbkA5E+3TMr82n0fx33GsKDP3MKBWMOcLGH0WEphJkgI3HuQErsImJUdXd788q6r\n",
    "M4shnLdpokHEGdThJBDmYkDr1OJ/P662JDn9l0HlzVHkKb1hkJ9xuLvPGM+tG/OwnzHW0aLtQfI1DEiUwYy716JJQchQaNBzS8jC\n",
    "BnTei33/xREhcruyiCVI3/5Kku3HosAWlmLLJbClwaBHH1QLOjbYDtRcm7bXlDBFYkq3v9mnvVEIJsttPDO2KPV464gJwZaeczKE\n",
    "eQagwOfCmSxhuvr8AM9sQMsqQr2Hs9hy7yvnbw3JlkDgnMZKN5lxV3PnEG8DlG4yQ6oGCLjEIL+yKdljBWeBkxAroWOF55RdKCQr\n",
    "qWOGOxHtDiqsxI5Jqf0pCUOQP5RhkN+NK91jxu4cK409ISgNguVsUxjRyUqDmCE+0rd7UiUvrFpy7FTEhMDlx0paWHFHGzcwK2Fh\n",
    "EhI4jbLSCWYIEYrsY+X0v0HNLS7Kzc/5/u1tdcqDv4KcNlfGMSEQ7jGhRARlROW85tXDTZESBialjqdGhUGP/hALmrJH6AVk5bif\n",
    "FCngZ1WO+xmW0+gMtHsvtzrK2Bq98smv0BwC21oI6h7UbeWSdnewWXnlJ4WBwoVBfkJCoMfhULnoud7+z/fhp7zv67vB/RpWjvUZ\n",
    "ytMSOD0oL/mMI0rgihcGPT4iW9CR6CK/f0Sx7asMfjzl+7ZBS6+lNtuM0VBblUn0cxL2P53ySk8QyTl436pGhCCIhpUTe0bd1E0c\n",
    "FSRQNE8rr/dsUk5xEiNMkKiG08rkCFpo07Fh+xuBHp60y5N+gMomHpcvr/oi4G3O5VUXk8oq6AFECH+Uaptx6SDv88vlMT8otBJX\n",
    "vJsWIsSm1S2hpc2KfnnSj8/mBI40CDQvLmnQ5T2XIxqix2GbkBAeES7X+0GY01+00g82BBUr8kYQW8PYTTuQMKIasvto5GWAID87\n",
    "j2I/rmzycmBCjz6uuo8DUgePZkutY7ulH92WO+Mbyo8pm8M6rCVe/4ey87lJp3jSkJQMYV2Wcs4WDHK7Cgjyt7clkQWN7fEJUyim\n",
    "sKWgExcGdf9RanTkIwrai6SSlLHPzXI+8gFsQY8+LBbE9cgWZkDKqOc9aHkataRmQXkzjUrqhjATe5D49g0LYo4PdJgQjFhIiJxg\n",
    "Qsrs2pluJcxFiCmhY0KILIW312shhUrcMFSLxA1jQ1lXPjk6qAmJoax9gh/ypMy1H+bq1VWh4kB1l+dFqFrKvC7mFmIIStkNA1IW\n",
    "Ps7PKHHl3VDWAr/pX0DZ/STDUgTkIMUgPx1wMqDzroTtB76N5sry5BVhDPK/ObMFHdlxZAvy4SgY1H11YkDr3CXfAVXfT9mCZkKW\n",
    "8JSOKfFCxMVQ+IhRiEyBb0OeYQ1TYQ3C3IWbIVCR7WLH3VJG3anvwsM2q24jDUVuU5yBFIci9HoHSoQsga7El28ENoTVvcHuGxOi\n",
    "pULEENZGz9sSAj1WfFHWM2NuQdpLDGoeVAxoRccEHhRMyNFXV0vIIKpXRNldPm/fP0DdgJb/bPiuASDv4RMZtroK7l1ITrYk4c1e\n",
    "NpMtlYHkLZkt6Ih7jBbMLLYhnZ7JiyDIe7ZwSX6PnJXdlLMHQ+VXcnlSHr8MAj2MNlcD+oTXPLeBCDQ3HduS/GjLyu7mrnObkg9S\n",
    "4oAUyd1Qlhvw8VEIRN8BDf/lw7awbkQJKckSeo9vpGJCkCJYClmCrMDaHxKC28dS2PZMb8C6irL/76WckZINZUXE9Khviq2DQPJO\n",
    "Kco4+7cyOUtphrKGT7w0l25rSbsVqgxLGAwmEQR6DKyqLGTcUS12zqpkQCvGJHBcSmVbaklgXwdBw7dPbNUoAabU/ARF10ikFkso\n",
    "dXcbS+ptCUv7FjDhQdCjy5sBfeNms9Tuys1gswFBfu9ehwHxHUD31lWAEEUrSEu2GbLbnzWy/cE59shjQopqYPsNtLyMPehxRNmc\n",
    "/dttqjM7y+d3fRyFEeixt2zZgFY8Egd74nbb8wxQSGhSRaDHpgKW5HcIrVpQBWliIchH5WKQP0q2ZkD3RPHaHYgQzQdNDR05ry+4\n",
    "tJjShgEtr0tYKiBQ4JSXnmwz8iY7IiREUY+4hhIR1FjL37qfhintx5RYcReluM1QhYFWDwSaCQkNSAwoJXAlFIK8Wi9Kr5sZQyTF\n",
    "OY4wIZr6lHY34xemnzleO5R6N+vgfKbd+hElhen7cC0pbpgapWfSzd23dEsRAY4uCPKLpRL6Fog2qxQkBLFvokQ/rjqrum4GBLnb\n",
    "ThD08IUqyZDruU57A1Uy4YwduIP03z5CLCHxxqKVoDfDGWZK/MdXKyvuZ7b06MSu9Lkrx8nGcafUuhUBsTtcKRXtblTs6YOU2KWm\n",
    "hDdWsbiufxDIWXBWOtoEDX7uJCDIr/lZKWc8TtE7mIWyksRmopNPLmwKwsQgITqcZaWnzXZXkDQqK6lMrghWikothsB0ToWmVATK\n",
    "vl+rAS2BPhgXWelsk1DoCBV8NRVIibPpZiXOTcoRNvdTSrzRxZRQos5KrptpWWg8/YZZ6W0z2UsGiduz0r5mSR0IIhDkz/ZZSWJC\n",
    "90XzqLMghWKK2I9i2WlfWWlqdy2BGAoJUUBKVoKcXKku3S4/X+paPkJy0jNAOF/q2gJ93oh7vOoFQT51eb4UtKO6sg18yJc4dpRb\n",
    "97/gJZXlFStRgK4IQf4kly+pLKtcK6F/EVIofOQrXwrZQWEQGJwv8Svf4SuxRyFfulaeUQ3zHEqRjUFC/imB3z2D+dLNDkLaPGaQ\n",
    "uRrCcs7V+LMRheJfQBmvykZpxwor4y2nGrT5BRAlR/cSISXOQJl5WMongJjDfXmWZChn6Lv9YATyUZFZyFZetok8ISV2uGZhQ1nn\n",
    "zU0tYltP/HQ7YFDx/aCs/LPtJt5O75BCMaU4ChDLIMhf98xSLSgBnRyDHt3TDOjUqKKhhChxjrEs3VDmlBl5OCAhirXMMiyB2rZR\n",
    "WRl/Oy/3+N0MBDnfQ85kQZzj656YEMjlOavR0U/Xiv8tEeix88pqPIxzlxGGAWNKvEDnbCjntOiaAkCP8ZDv8TDjMBLIlp1zNSCq\n",
    "3/j1ESWeb3KzTTnOlu5H6Aa0crn7sxoCPVYjCHLhgTkP+xnHJfe2BxXrxM0lueokFg1zuY14PeKVT239B4TI458LG8KZD9oaSBED\n",
    "Os+97zuUkm0zEj1DV3NRZsf3rRnag+g7IL+3LtWAzj1wZI2QEm8jSrNNuRIYmaYgkHdblG5BbXtjDFPinXYZjgLcLLkmCzouQsgW\n",
    "RI+SyIJ2r1hAQiSb56oMWs5NQyAN5CqGcObUin7kmm0dkuPLN7kWQ5h3Ah5rS6221NTibGC5NltqoWc2QwwqHtRtHzQ+JqYdyMfE\n",
    "5KrsKZ/38OKDP6LEGnNuyVLq5jWf3MgQVpIN96hjbmxLReIWBD0cNE1sddK2h4mmbKq8ZJrKrRgQJX56wTGo+W+tFlQZHEsQ6LEH\n",
    "hqDi29Ts1+VtDtDclLlduWU4PPa1YSnHq4f0U0q88erJUFbuI7/x6srg5pVGed5IwyB/gOxsQEuTH6FDFFFilRBT4rNWF0tJm0zh\n",
    "mDAiQraEY551/VNcqW3rl+3KZPsdbRpSmqGsLVC0cEFCdLzq3RBWQLDLu5W7Mtdxby/ed5AjGcI6tAURU3mQJUh66lIY5Nf6wRZU\n",
    "9rsnSIkdVZASG+647XDFFbWNHY5sCde8Yj63GNDpBQkbgiixHY5qKZ8pw2eLgqBne5v9qHYdkN8rB5RYu87jtuoZodO3yfIxheOG\n",
    "IUos54xhKeNIt5N+yIlXsZKSoaxA4nBzUxJZSgcew5LYgijvzB5S4rcGS1Ijhc/rc2TnpJKyAaVadhn8SiqWIiBpSEnVgo7XdUtU\n",
    "LqDE6ddKaoZyZncpP6WEvnFIoQ1F2bToh0RfZqmShiFsX9WBhOjAUSjZJtFWycGUcPxDSjzLFCL7JeMbFLYU2moihcRRapxeqpAa\n",
    "Gfl4BS1617pQsYRG2wlGiY98PgTt95AY5LasGCS+JGXI5XQxhJ8ECFEcbVHy5k14NEOZbtWvkm9BfiJVEuYC8fHIjwEp+5qPwfZ4\n",
    "xwAJUaA1JkS2pfTOlWRlk5yiKO1z5U8BqgcEPWxACZbc7pS+7/MGJOSorYAQXVouShLV72a5dgOQV9eLEj4niLYn1aKEz/lT9AKM\n",
    "UUmSq9z+lIow6NFCZdb9XJSzdVNgUHUrt9ImZxKZ423KaO1DFAplgKKUytmUxNsVVimVK38OyM0NQd4fVZQ2Odt7ZVUMKkeU8EBZ\n",
    "lFK5cuWM54U2CPJiQ1Ha5J3XJ95MKRFyhoAx7cKIi9IXZ5hZ3z5vgClhkBWmxEusEiUnBaU5KUqInKAsm52RUhhXVN3mqUhMyBHh\n",
    "tvEV6YUWWCUerjA8BvMPAj3mCSUXrqg6kEwegrx0UpRAOIP4uD1P8kUJhJLuZNKuTRW0KXBNFKX1yZWx36UsL0qhm91x3TR8LfU0\n",
    "jrJCnY7To2srAjmfZbnUugMksj26XfLbQZmuwcDpUy75rRzvmoHUBhjkT0uXRneAGlq5L12uHJl5chzMjQnRPuaS4oqKsIrPpZCy\n",
    "6edqm9XqU20tl8p2gsr2MAgpHFP6TZmqTYovikAChb/BMIS5fsZiQqnJUiQ9U0eWSgZ0PoBhB0FlVxJvO7EqS7yeOAjUsVKzIZyP\n",
    "tdpfsxYLSh1sX2t1JYEQtFKbAfH13FTwSeo3K8fJL/KiljoM4X7UTTekJVdqf16vL41sSS1v9vlN/WJXTJLfCiPQozebGNDyMQWv\n",
    "VpaWDWFllvILXCuu6u2Lm5gSzxBNWUJ7eUWktGZAa4X0q3brFnQnk32vfNjK036v1JOhnFkU7K/WlS2ozay1qs4GtF5oiWfgrn7o\n",
    "T/SRgMxdpd8/ro0SerOGXixBwItBpVcDggFKEOTDjHFJ/hTWm20TerQOgh7bnd4NaL43GMXbQUK4Ce3DEcZ2oIzbilb4z1q/Ijfj\n",
    "IEvp2zhSTIkn0MGWkrZvh5YhlnK8NdB+SNl9SzaUlfwi/ZgSpxPBlHhMjmIonEp8caaMaglcN+MTEmpEUOOCbyXartOjG5C6kv3+\n",
    "qcrQzX1hVW5NCYBC3zOkxKOhJjKUbfRQTWzrEBAQCkE+20FNynjntmTzRnRN2RBmakofiQJBfuqsqVgQCfqIakEVTJ01KRMpd3rC\n",
    "sMu7oSznYIm6HBFy1FHDNor6c1eIQW5trZRs1Ucg3OuQwYRgL1yJbDNSil2lmDAigjLGel71Ds7pkBA3SQxhWfKjKwHI3y/CILeF\n",
    "qJRt+3J9RidWKga0DjthLDKmhKFBlaqlHJNftU1RY6SdsVBhwqxK3VAIOY8xyO3iKikz79/IzQAJ0cXmyskSGghkgyB6gJRpjzNJ\n",
    "TLjtr8yGckY+2T5gseUm3pcLKLGPFVPC59QqZ0uRzdtclW8TXU/TbO8NVb5NdMYw1Pr0E1VuBrR93qJyt6XmzbsFkBBdpa48bJNq\n",
    "2jUJEaJpS5IloBduqtw2OSMiyvaJqCpsKKmNp08bg6qvXCwoE5hYJdsWSosj1aoUW2pvu9A/SIlj6atUSxngZe0qyu7OXIy7pnRL\n",
    "KePpycAgJxVVGRY0hWc/f2ZlJDPb4X5RQJRYMquZDGV5L7zJQJD/8swGdL4LKnvQcMtWFlvdoU2GX44ooeuk5myb0vjpucQgv1/N\n",
    "xXZ7q2DTjUA+5Lvm6jqanhnNam62TXcajfdvVVZ53fr0O2olVfDc7YJA7aqkCj5zoXtHFgY5d2NVCsZKtCOoJGUvc0coTwcJBD2+\n",
    "TqkWM1kMeli3KhFivZzT4qDXqhSF9c5PifNJVaUnrFd0QMAkBj0+qLmvvl6SejUDRIkjDjAlxbV0S5mODn8IUlLDSnuDJmkI8quQ\n",
    "UiBmIEkqm1OdUiNmqbw7KUFCEImECcFD8VVpH9zuROBBk8QScnlmc8Sg4kHZtnWGAXiDRKDHAqB0kxl1MNAko3STmZ8njfgucFUa\n",
    "yh2g4Q0Fgh5f2l37tinvMSUMDK1Kj7nDYVrkTKxKneFhdIqXPlFKDV/vntb3q2pViTB3sINfVyDI2xQE+f2CknPkytBO0QcBQtwD\n",
    "2RJaj8X6qoSdGQKwv7sBKXFAaVVa0KSg5OJVaUETRCARcVVa0Eq8UzenFkgIbvxiQjTnKK1pEgit6gj0OBZfElQ90thQHPRSLznq\n",
    "IHSQvBaCHtuwS7M6QHdm8tcf9dKsDkoDCiYEeUGyXsJWVZETft9/iVkHqGyzE2FK7LG+NKx6vE5VgKcSgvy0emlYC0RHWLtdpC9J\n",
    "6ijpuhK9A/lLWPXSqo7qeDcwhjI2OR+u9h+BQI/feJABnZlRI3faYEM53SaucrHlZnRygiD5Rkn+zk4d2bapoy38KK6kEl88hoQo\n",
    "KXEd1TWDNvu20WwNhcDZcSgbyjqZ32upwxDW1sv1a0vJllr7c2sDQd7T2JKyoaK39i/tg4QoI0JLbAjM28kAUyimiG3WlVn5tVmA\n",
    "QCEhW0Iqu1AMTMk/p4RXB1sqhrI0No6+BBAiLaelahuFPFgYVLzhApAP32xJDax636umPehh3Qj0GCfdgM5p3oGGAd2R/1uQm7Ub\n",
    "JVtdqrF3FhJo/JAQzTmNyHZT453zEVI2o4HYUnKPTwmN1BC9IoM2zcqGsk0Q16jYOho4qTWqz4Y85mRSttbPtO0lqhoQop1Do24I\n",
    "KqPoe38gSo8pw1DuFJuvzUKE4LDUONlG0SYpa2MCBPdITGNlXucLFPElT0iJH/BpLIZy5nimH1JoQ8mWgqK0Gxfbeto83d74NuMZ\n",
    "Jzb4GZYGQQ9b52ZA54N5wU/ebamS0AchUPOgYUCnC8TagiQHAlcDm9xWNWOmjmRakcFASihoNLltbD1eRmcWzXeKGMoKJo16V7Kt\n",
    "o2/fL4SUzZQvt73NWJ6WDqHutVnVEYDbGIIeOwJYkl99pTkQbbY1kBDcpmnSHYGfh+omykDzt4L6MSX+MXIylPMmth0rCOSTkbVM\n",
    "tvKewUYKgoYvid1npN1DiC2LpVQ00eVsQfPIEbivISHy57SsTLucoQGBF7jlagjridvk291sqZI361xW9lXPa9aP3wuAvMez5WFA\n",
    "y1MSrQ0lWcI+FVUrZBvSZRdA2oqykJnwkp8uu1bEgJae6HfRCOQl61ayra7sNhqlWIJsc0u0oiyh33c+wl5ohrJuw4U/Trd1jBpf\n",
    "XoKElXThjaDsZRwxt14jazUZUGKJPSStkit1NwAq2xquKx+mGWJAMDFiq9mWdOXZeP9ZarHl1gHmOAh6dNRtE5zu9BLBdzdD+EZ+\n",
    "4la7oaTjEfB3m6jDNirBbkMg54trLVnQcdKxfdDu337GFCXapQppjQ1lXrnzqepakyfIh/W2lm3l+4yXmBI7XloxFEYPYUGQD+TE\n",
    "Jfkpr1X75WXEV7FaU1Y1c8IUMJpaN6D1/mOLuwpQdl01LKUCraYpGWZFgrXvgvyOAYEerk+lz8zMNAk81wdBj5OEEnHm1+VNirSm\n",
    "1JlV6n68K61mPVjG24OnkmFms1raLtBKYblyw4R7fKW2TIKM/ZcgSrzIKiWGs/HN/4hCG8owFBppM8AQIfzdlawzCbVuVlBIiHx/\n",
    "ShOaBEIbdQhy+RiakoomiPkp8GCQ34wqFWhGkc2cJv0boMfZSklFfB0X/DyplJ/1q4B3h5tSe1bgH2+8XErtmTFkx4srdkqBID/v\n",
    "KEloxiOVsh0yShTiazsfqsSQEm/ouxKM5vtgiXcRXF2pQZPC9bkadyXnzLgfznG2LUxIEUEsoZ8/TtByZUpv+ey7Ukpm8EzbhF5D\n",
    "QhRp0pUgMrOesMR5EyAhkg9wDRI1qdlvOB6VCmroroYeb0UhIYVNUsY8zgyh/T3BW1cSyAphkqemg0HyDZB/R7krNWS2jzfaMSZQ\n",
    "RGBLyOlwe75bOKLE/k1MCffNXYk0PO7Hul3/ZPe5vMu4CSkUaqddaTmTQgSGNAQN3957zKwH3kBm537pN22lPGHZ3VKGlNhD1i8F\n",
    "56DUTQBBv+SYRVgZAd0uul8STFMPKYWxZP0SYZqK5fAdB0FOqemX7LJAaYx47YWEcFa9JJe2giimFuSN8hJZDtBgMEdcIku7M3Rs\n",
    "RsSloByUCjRuCPLbnc7KXE63pL8p2Lkb0FLzXb6zzsOC+m6+kWQJR848+0MikPcWdFEm8dmItPXirSuJLUh6rIV2EUuguhM2MUV+\n",
    "TKF4mEi2lE5HDvqAUgzlVNNcN1Zb7nQ8bT64uXLBUblLt+VyeT7h12UY0LpFFB6mela2caUhCcN+eiZDSUW2P1BWdtMPL9Na417s\n",
    "JoshnHmgbF/nbEBzI/zYGORiqz6CPZoFVQvKLY4ygoRwa5TvH/l+EigaOYgQqV6YMKI+7pZwP9P4RhiWMKOCwiTWmBK63Xu5LZKu\n",
    "9Tr6cESgwBnSCxnC2tXFZy5Iifd15bb5mYxi9wZLL2II9wu/QR3Z1kG8OWWUYgnXcSqoA1HiX71U+yVp2yxlu3JmsnORa710C0Kv\n",
    "n/QyDIjuN2reW4sooWu1V2VyZ7joY++AQI8tbCUL2sdQ9sqWUvP2E6syrXKYL+XQfGs2lLWpcS4jCHpM0xDEvh+KBSXw4hQGPXq0\n",
    "2m9ttNmFIkIUGNVrM4TztbHwB1C2ez1NGE0GdVhC689QCgiaDzdqUEsGxAlcwsIgd7u7N2Wr7fSaeJNobEDM/XmTozdxJdWnLo9B\n",
    "7EHZgNYiFm/9WnHl0nbL0+qzKQ8Tbsos+q11v9sRJPD7k5G9KSu68soGbmZICF16bRhCOnKARRNFT7aOmaPZPcGAQcWDbvta8uoA\n",
    "vdzZgFY4g7eKLgaUjlsP746ofpvRkkE3STIhIXTX9dvq+Hz6LgoPhoQoOgYTov1cv636o/AkdBu092ZBaZOECxI48gr3bgjrgoZE\n",
    "tyV7v8103WUH7yn2kQwocQUuQQhyjxr1Qba6yptxN5SRljM0MHYgIcpmpzzEUNaqHk04QxltvSND3nfKoxjC7TkIGlUtZesyHcpe\n",
    "2vnGULTfH90ROvpdhwHNO+DRiW6kZAjrFE4/JER5h3ANI6qB7Dd0OWMz3gjKhvr90pgx5pHEgJb45RyPGOSyXIykzGlm5QRJQEa6\n",
    "TWil5gbRDSPdRjOvf1aJz0ojNUO4HaOmVAQiDzqtqR9+WI6PzeMSOfrh6W3ogwDIP8E3Ls2ir4uLheOpdVxyRD8uMdIp+78Nx3FJ\n",
    "BZrinEvjEgcWaN2JeLQ2W5C02MONCYHLY1w+/6M/Ej1jcMbl8+/LdT36U+AdlwO/H37z7cPOmFJiSndNAVkJBw0DuvNI7UAPc2Zl\n",
    "LkpV5z3ISduDlR0VHXf/9ssgQrTbwTWUiMCW0GosTg5WNlv13QbzodmAUgUP7GBQd7MmFwdaHoqgfYgQTd5cDWFlyxz+g5ottchm\n",
    "EHK3paY1XILHOQcPSzlSZL6PW1E21++dqu1lCPKTsSizGeel6zCF0xA2FDpeISs/pVBMEUNZ2c9y1Ce3Ua2rN+BFxiHFgJZaTNaj\n",
    "NaTaku4IkbeqESFaTKUZwvK+hf4MTKGY0i2l5ed1dQzyo1puA50OXwGi18jJgO5Ei+8tzGQoy3MU/ciQUH5ICJfEzJZw5+R+/wqx\n",
    "lELx/c2Rlanyi9Q+cjGgdLysJ1vQY63L1VbX0TKWlX3lO7jlGyC/amZlUeVMSegnpzwM6BSJ+h7kFNJRlNnNyVPAxg+BHnNhIQNa\n",
    "kXbRigMJkXEVNoT0DeOClDD2bBRxlK1+gCmhSjFKtl0qJb5VNoqy1XaHOluDLtWAvvEO8yjNUPhQKcZ7Q7ptyEAHJAiq3vaGbe2d\n",
    "Aual6posoZUjXkuXikA+SBuX1NxKVpWBjjMTkPM3jcoWJNuY61Fvc5nePgYXEiDI68CjZgNaifyqbyEC+QN0LQ4EbmiMepvXSuia\n",
    "d8FTozZDuc+MplwE8lNs7Qa03gVi30IAekyx9ba7Oz+rL6klA1rp/fzKjkCPXoMgv7VqZEHC4CwJQX6VaWxAyyvtf+4mrk38zEE0\n",
    "mjKc6RkdYNpvxYFqHFQ7WjUEuvwRPyJEfqimzOlMLxolbYGEcCJq3RDW4SBUcUZTRlfO+fvRmQDkdeXRkwXxfoHqynDqiwsSgnyK\n",
    "xNHZgo4rgO9d1cUQzvdo7GQHQX5EItCzfdmCBoMNOQT5HWRXhn0pdaHMjinxBqVXQ1lifuRH6M0SjkuA0WGwd0uhvBMQMSU+YnVl\n",
    "u1dezDCJwRjJUNZb1k5EH0OZ5Xnl0scuQNDDyzbYgOhOmvXewtvmrocaU3kX5jAhv2ukY2RL2GUtwYQaEYohrNmtB54dSOCohuo+\n",
    "ehN5iwnhRzdLSOBC0Bi30QodqeTjwJgxhqUc2x5nXAhkj4mUUrKg+lwsX0DWp0ApkQPRmY0Vds4Lob7aHKXEhjCTlAWZfl8I7xZB\n",
    "KYkh8PWErvlQAHKhF5Qu0WYcysTuJiCmhGGllC7V56BQHB9J6VKAFoGP/LcS1QEooduK0qUIHc1Kuxw/mBJmc6J0SUrj0HXosRl8\n",
    "AQ3/Yw0D4hSngMOEYKtG6dKgFuHOVqubcelOByg/b9O/gGyAO6VLjhpLZblePXxtHyKUiCCGkI4oiPYzAgXTAyJQ2MfZEOZF/I3V\n",
    "IUraUIpt1pE93P1ICFT8z10tqJeNFPZCicciqYE1b7eWzf4GU8IFiRJ1S+m02UK+UEpMGYay0nOy71YEeowiBCoOxGrQzgsd7bHP\n",
    "ewH5BZTJgnJ7HLgxyMX9vZTkl2JmCwJOGEosFpTWCadaUHYl7SY6RFh3T98Iaqi0Q/IhP54g6PFLVANac1iO2tosgerjlS5K3C0o\n",
    "DTDeeRjQmYvwvWpRVjNfcJWH3wSDHrsRIQNiKcC0hA1oFy1ESZRxjNONR8FWS7IhrLvBOVhCpViC8HaDJPcPTGeMS3A7GBPC/Z80\n",
    "Q1ge8RF9BSAE901eaoiMVLr9hlwfXj8MmuqsAd1Guh4fTceAf60aECiympwMYUX5DTc5QZCfMPNt1VO9Gh3M9BDkh2ZmA/oM1dj0\n",
    "s1hC62D9gKDu23ePiXmTApz+XkCPjygGdL9c8D5YIEXCnT6kxBuSXA3lfqzUtF7ZfD5VfY6bAijhJZ0XSrzRyd1QKPF2+5eHrQUE\n",
    "41EqyYLGOB+cfbE5SIiMtJBteZHN5rioYVDOIMcc+N8oFWXj9VzKq/9UBGrGT0qpZAfix7UHDCIrDlMqxYJSAoMTgdgf0Es1IHRf\n",
    "4/OC+w1q96Vo3oIee8nSDSgdz7cEJgEIFC0UZdhmSD/zQr7+wjVZSkvbE1Ml26zRgPMDgvg7IN+3lcFHfQfkBM/Ps34WxHFsHSaE\n",
    "e7qqDLifAR412PrW4ghoJazVgJiej859HnYxoDQKcPVBkN9z1u5A8KdTpjbOgBxfUksWVNaaZWeGRgaUKoWq2guhBD9ju41jKsRT\n",
    "/PUTNgLN2dGAxIBSjyNzXwgt+rhsCOdL7PIjQng8a8U2KT+z+2PQ49zbbruconoeYdrOT9Y7S+B6ZOJ4JXRDWGH2fl5tw4Bm6Eh4\n",
    "S/OTP8c2ZHNHHBNS1HJIkKBv+m3V94Ou0cmmK4OV88mlGs7cXSylLHkk2gF0ZV5ZJ8l8bRYghJNsL4aw3L5+dPZqSx3nhcjXUpsl\n",
    "HJ/a3wndEvozDu5z2cW2VTo4jg/105f7xdZoFYaU2BE2yFCWvUTnVEgogfd1sCUM2igLQ5lXPa4eBmmpMSFJ1KRsCOsb/NIEQfQd\n",
    "UPW/ZPlOdQjkp8pR7ZdSC9NyfAIqbsIpyJP8kBB6FSAh8vePbghrqxg2adgarkeZXq2akhoI/bzJ8h6K+ULoEYEMYell7zfzXgiB\n",
    "B4kSG8ISc39KCLJZvxByVINYwnimf/ocbG1XUg4fKPycN27CuMN06zuhWgLT3iCaoayAF/Et77bcLrv+QASKCMM2g+J3hz8MSzjO\n",
    "VcGMDimxh4eUEries+zxAYWUyCfXvs5t20nJbnLtQaPuhISoO5XoJurOBv+MEI0pJb/Nnulg6sag4Xujuj7meJ4kpZ1NAnjL9QVU\n",
    "PKjbXwu3D4GyL+mwX0rr7uX1sM8O5A9+dIpdJ6htts50ylUnQfIm1SHRKV59KFci6uS++xSvDlA6khOXPcjGXROdgtUBWm6naOI7\n",
    "taiTQBtJjE5d6kOQW0aMhjakjJ9Twl0knfrXh/LZ0NLzGSIMcokgiE6R7AAtF1g8t0NK/TElvGtExMNQmNH4gKDuPlGU6Zf7ybtg\n",
    "Yw8pFH8irKWGv6KQoawIXfk5pcTfwpbSx9OFSyIWJCvX9PueABHWk2NvhGwIK2w78BeRFEtgoKGSqLFTj4BOLy9DUMq+B5oBUeuP\n",
    "e0lE0i1IOtjWQJCfj0VZbjtD0R6/CgB5/zTlZEDnhWYHIgNaFx6c6oVB7Tsgv0xmtg1POT43QULkcKMsjtCfzlXK2ba11TNO8XW0\n",
    "ZGV1/X60x/4uEFR95dWAlsNOfEkA9JjIIejRpmZBt+721oPKTMftJ3ClAhD5/U4eBrQC28IIQCrJUXY7eUSIPEm4hmjrWch+atu4\n",
    "RancNkvXrSpvgkUMiAoIfMIgP85LtqBDUJEt6GGZpdiGV6CTY1D3oNsy6XoaPvCvUmmGcKb9CDq5GwJ/7pKRn7jKsM1oW8UcU+JN\n",
    "V0227bR7BfmFEh8n622GdG5SHyeMqkxPztjY5EFiQDNKe7OJgZR4w1uzpZSx3UAiShwiQLW4Dx7ADGq1oNo3Q74qa8z38zV2SEGQ\n",
    "czlS7RZECRzQ6zCgz/IUH58RIQouoJYM4Z6SXglka5AClnwIclIuNWWS5QyvIF8SAvlJpYkBrb1A5Mdoyp7q/VJlNPxbsRQiMMoQ\n",
    "6LGCt2pA65P8vrE1W9JUV+JTYVMWdSbx8kottWFBrQOzQ6DHzrYnA1r3WONhDCnxfNiVJXUdrG6awgaU4G8DQX7XC0HVuU+6WJBU\n",
    "0Ic9O1D8pBJRV7Yzzl/m0b5qQOtS+HC+kN4MaMW6DhtdQ73b6q54qB2I/OagDwvKsj3HjtsQlupcwLcOsqBBu2h2TBk/p2xaz4Zy\n",
    "z+8/pMTL5BBDOcOffkrZOE1GtpQGLqtgkJ/OR7EdSWNj8KNawvFUXdiNzXYjp40INLpteeZnzCIG+YVo3Fa+EgsMZLII9J2SvNOT\n",
    "kxoh8x69xK5KTuQIPUyGiglRFDcnNoSU6y7okpMySNGXrcynApC/FcApWxA9HybAIG/MnIoBJU67oYgpNaZUQ1mnBfYfhUDk26ts\n",
    "/rrIv6m8G8qZTduVi0AugJvTeIL8yGdKBqRub722kJTxlTM+3TkoMKhZnz4TG9B5eZ/2oOw/A4H8T0Zi29TS87jMlG1JCXiimYoB\n",
    "3bkcd6DHXAFLelRXbZt6/F4IMSm7m7kMyvPOEAT5UxaTsrSZdDU9RWkIeowFJVzNyITjd7Rmo4SrFb4wdoosK+lqUo43xootV9mH\n",
    "ehLvfcKEhBYRxBKoxQ4lVoLWUqvja/DEStCaenh9JoPDIL/rY6VBTdAAwfEQ5M9DrFSnW3LvHnSbh6gNVN6D/FygFKAl825/RkCI\n",
    "IvxZCTsr9e5y+9h5C4H8aQ2XJG4MKoVngo4HPYPDLKbEQ0SJQvMGPz1z9hArjWeCJG02IUrjkeuOupOrMchvEC7xh9Rt93iZhJRN\n",
    "L1RDSTk9d4F8CUMLtGK9N01BlBE3BVF6/LP3m8JnYha/mkOQ30Qh0IwIMqBhQOmIj7Dj4BKjDlBuYEa6JCRS2ZSDGEi+1CG67+BH\n",
    "wj0kBDmriS85iZYGX9uxFL13P6SEF6cxZcS1ZEs5UrcFfVUs4Zg1a1RHNZQ1LwUBYZybJVw+6qAOZWL5TBQY23cehrJedA/yDnBJ\n",
    "hnDeCu5BHUWZ4tyUFmCvhS3oCh9+bQgiRBNnEUNIDC4lcskOBGJfIcg9HENclI1Mn2WO70NwqZbAslNLMCV0eHBphkIFnNwxyK+0\n",
    "RVlbO2PyokW+DEs43m10pQLQY2dekwUVINNB0GMNrGRA6xJvPMoQZeZLDihsm5IJHFyqsk/1bsK7wdRsCOsRGn9IqMWAPhf7dtNh\n",
    "rbbcPLaTLqTEHoKqbGycPowUNwxQdrV0Q1khNP5AXIctV7ZuNW7JlVsfaXiJ220pdL4S9JhxEMjfOeLGBrTevwmvInITW67k00xf\n",
    "zAkS3rOivRBqsHa0bAh8JBx43zG023jXFXs+F4X3z0aUeEBDSuyla9VQ0pHo105ACORjRnFJj5+/2Rb2OKcucbvNfl3rl50WBimb\n",
    "rX27x83UglsGMywE5e+A/NLQkwFN/Te6ZMGdbKlDjhXKlMoGRMeLItF0Dimf02pAEdv21MF80bMDAV84d2Wtl8A9vgPK3wF5I+7V\n",
    "gJg6mJ4Q6LFr6s1WN0uKz3iIshmZvRvKCvne1DIsJZXNTnIk26zrqvArgSyh8e7yOQ9lY+V897xHdYgliAC/88gOxLHiBAkUEool\n",
    "0Oa+Aya0YFSPagh8vYJrPrQZ0Dpt92BtgoQg1pZHt4RWwZQFQV4xGMP18fMdV5KkbK6evotg7cSE4EAliSyh0fm4zJuRSmJD+cYt\n",
    "NkliKSz7WhCFf06JnpolSdlQdg+jkaRi69jdMsWE8BdRlv7ZVdf6nOolNQM6Pbi8B2VfUjegpeY8qhu2TfJ8//gF5JKVCCkDHWew\n",
    "ultZIMjHXAiRAdEAcSoY9KiODeh8f4v3IPEliW24UBxWJpRt+1qcAZiEivugEt+lFLrNacUk8PP8LdQM6NRHHOi2FFaOVNqDyndA\n",
    "LkRfaBgQH4c42yZOBrTyw4YLP6TEZ2NhcpT2nNAxyI82ZvvlReKbrcK3Rc3oA5Jz+/ve2mwopzm/LxpcDGE5VvxvBkHD/xwA9Bi4\n",
    "rCxSzmdKXBIUDGoe1Bxok91QuFsCxa8DvBCi6Z6HIaS5njq5TyQZ0NpKRqUKWUIFQWgibNsqaXduFlHWdUW1urOSSDYgan0XtiBK\n",
    "MbrDITZNQZR4i6BEptl62kanihKD+MpxGsgXogSc9ULDJqevKDFnxg8wyP4mSsyZpV6p5bcgJ7aLUnxmHoXedoclURLOjEVIdXc5\n",
    "DlI206xSZGbDrgcUTOuzLTej0YJAj4lViTP3C7exxUHKiD+qGsoKVt30g7K4ft+/eh/riBDdx8E1RFsCpRjNBA4dJDyCoKcdDwOi\n",
    "Kzf8e38oKYn77XCsEYXsB1LeCR6QEj4SSKKkpfnB9x2Hd4pYSgYx/hDkAxhEiUsrogVoG6LEpVkSil8SpQ6t/BAShyuKkn1WloW+\n",
    "WRMhITqUKs1oPdZcwOqlJKB5a/+4yPw+0SLCyt30QlDS0CSk8bzaA0GPs4ASg2aky0C/OwI91lalEc3qCj29xRDkHW2ihKT1lkje\n",
    "nsoRJVZAcC1hKie5BCteoSACgo0wqHpQMaDPJYTniQCBfLimXMrVUR2XOBmZXCLUUeqVrTP4bkTpce92S7nsKqCMm8JnKsQS14Io\n",
    "8XR6CV+LslJzR+faS986CFLDl2MxIYU1sCU0BqMZgvxovhStBVoR4bHbCFE2W6iWLaW2neYsrdjWlx5rw9KqJaS6r0NZ6edkkdHZ\n",
    "r3UD4n0SZEyRuFOHoSSuYHhDkF8De3KgDHweEETfKcnFX0tXtptf4oqlswEtieLRJjGgddkoOtD2bEsl3mWrlV6edYT7y15tHbXG\n",
    "N4shIXRq9WY7MHXgg+zdlsoCDhQQ5G2oD9sDLe16ABHqTwnRyB3Kzsp5ZySe+SGFwhE2yFLamQT6Z5T4arYMNpSVgMNpUjLEgJal\n",
    "9++A/Fo+1BCo92MU7z8OJNAPCVFicRnFENLujpGMamv4xjkKUTYjfzT3Hdu4SkwpsZF1Q1nvLriroTKGA23TFuWUbKfWfmREUOXm\n",
    "RK7n03NXkJMy0HZ6asWXBEDPksSA5uoZO61yypYytrsvSImtI6diKRW8wINB7EHVfWJ6LpMQ5G/S5tRsj86DcDDuMKFEhG4JA8z+\n",
    "OQ1XKvphKdkPOtJuBsEemchRxu56QSZrY3ysVu8fCAn1Xa/OpAz0epYzSB+aKRvCnTnL9E6xpQo/T2sY5Dz4maqt7kps99o+ROAf\n",
    "EiJvRaZmCGvLN/zHdftxnJ7+mkzK0saZUHZEk2fmZCjEm9zvmBCkR89MtlGUdolIMrOhrOhvdz8BgtxL6JRZbOV9XTd4//U4W0Iq\n",
    "8UYqc7GEsQnrwYQcEW6DmlGZGexDMcgdCTI3C6rg2bzM3YBOASL4gQEhCpjLPGwzjsybQRL0LMnWQT2+GwYJ4e8oZAjnkSBa8xAl\n",
    "Tn6Thd3PRGBxgyB3OS+L2MqJ42ujWbJrbXpKHxjkfMdZiu3b4/pvNM2IMk/6hqoFCfHnNUNY7xsFXuUs3RJQavEswzbjCNsMPxVR\n",
    "WqQ35JxsU6SDzRMCeR9DzmQrJwhS9sXHnZHYhZezGMp508+1MNty29isJYgQLsa5GIKKUH5vebV19PMFa9NyBPJmn5sFjQGmWAh6\n",
    "/ADdgo64saCjlE3JeYCIpr+SDGE+7PeYZwrZUvNWj8yFbbm0W7yKWEJLsS8nl+xa3o5gjVdCeRLCzkSE6EJBLtX2k/RnJBYGNauw\n",
    "59Js1XUTYZ+LspR8xu37AwYCPfaGZRgQHcEaRtzPNVlQkzhOJFeyVZd2LgVvBDYETiCPEwa5wK9cBYDGN0A+6BWX5KLDshKMZtfU\n",
    "FsvqWelCK/eZoKqrA4GkoRjkJx6l/szw3rJ5JCEr7ecKGo5iriEhelU2K91nEnLedRkiROanZKIZ60pAqMxK6ZmgVp7vSmGQH75K\n",
    "3Vmv46HjvpJqVptAHAoE+UQwWSk4K5J3I1Flpd9Qu2+uvf9CSr+Z+eRSAXtCpdisN+I2rwlDQrj3VtrNemaPNzOyEmWuN/N8rEhW\n",
    "Ugqd7xU/jAOC/C5Q6S3rFTqgA2PQd0ryGn1Wys1sk5TtiVnpOPOX5wp8TEq7WeHZAtYxBHrsHJSks0DgzhkE8aNN1YDODWv4rc1S\n",
    "Ej2TIGaly0wQ8+ZQqhSUGed8HBjdLwNAPtEyLsllRMlKTJmB2x0tcUoLoSuvs7c7pX6s6mjr2VYyx3oTb6OuZyVIrJjutJmJECGK\n",
    "Gs9KweB0v3P6/nspOYLTfTE/OsQgyubco+SImdeO8u66YlZyxKQIeOAlKzniStz3OKdBkBtARakU87es9ZmwqyiVYoKOGLjXnwMT\n",
    "KCKw+/3qM3a0KPViRqzLJu8SJpSIkA3hfo7HNAOAnv2KSqq+X4ttX0/PwyUGuZiQogSNWV0d6GdstqQMXqPBoOy/rhvQ2heHdg0p\n",
    "sShUlPIxm5I6ai8CuT1IUfLIrJx4l9O2KHlk3kqQ/HRlFaVvrFyLtJtFMCXHFDGUVIDfFYL8wluUWjIr7/W5p8EgPwiUWjKrayAX\n",
    "Fga5XXFRgsesLhHqaADy4kxRSsisroKVsyglZIFanIATEsK5TMkofEY1+RvLEOR3IUWJJeuFUInjODAhaqtSV1YGTIlDXzAhOGMV\n",
    "pbPwGVgVB20Wpbqs9Jk1dgRCQpTJrSiZZn5FrbGQCQlxxxZLSOW4fvQTQqR8FiXZTEIHGwUI8huFonSd6zeNDlyQEMVsF6XxzB9I\n",
    "+i6QpyiVZ9lN20WwQkp866MonYevu/thQASmcFyLGjTz2nbZJU2ClPhggymbWthQzidErXkoqWhlGG27aMqidKD7qVp3wMMgd0W+\n",
    "KLGILzdYZGqIEG4SlVQ0r+tc0njwec1Sjth/123dgOgIZQi7DVA22xQlF3E9ZWy/8VPazwRxfqZ8xSC/BVcC0WxhBVePMcjPTEpF\n",
    "Wt/aY/9hURrS9RitdxpAkPcPF6UbrWyzjEpCIPIfAUCPDbxSnbi/ZEIsSkKaoLSNvi9KK5qUlmL5CxLCnYwSmviKGolNWElN82JQ\n",
    "6qDblLx0XwyKFhxIiGYBpU2th2rRr4dA3qFVlGQ1bw8xkACKkqlWPlze3fiClPjNjKK0rXkv6ZAE3n8/JVVNAoGon6KUpwUaOwdQ\n",
    "UToUnz6jxwkBgvxZQ2lPkvQT5VuQ3ysrgWpdOhq7tJJFyVWS7qcOc0QhW0uXZ6a2ovSnCWore+H7dhQRwg2YkpxWQuQG+lZJTuup\n",
    "4LRzPhUlOq07YPUZSVCUniTXYw0j7jZAIY5/HFRLGGFelIY1P1gKMEsE8vetcEmPflBmSS/BOBjk3RNKuJqJkBMQhooSq4TMvvPF\n",
    "SJQmtdIr01PlKEqTkuuOWjSzKulpEkaNUz5hQuBKL0q3ErpP6LbLlFY1QaXtLsxiShivUJTANdtewk3K/weV3t2F\n",
    "\"\"\",\n",
    "    \"products.csv\": \"\"\"\n",
    "eNp9k01qG0EQhfc6RR+gDNU9P5q5hW9gjCyCF4mM4iyym7EMCligeBUwgRDH4GwCI1tjj35mfIVXNwrV4/hPIsvu+V6/qlc1R8PB\n",
    "wafe8d7hAX3Yf9+n3v5x/91g+JmOhoe9PvUGH487u2wJfyTHg2QoJZcczY7kWPvDxFiDJW4JF2jkFCXZMGJKutzZZedvsZIR1pih\n",
    "kqlxEZuXdJwyBYnCgbqcokKDh61oN2UKA0VDwrlkKHCPCisPb1ThUqaufzh6ujWYocS9ov6MCjWKTSPHFFqVxoRrUatazrAwMkaB\n",
    "hbHc8td6oiBispHSXcI3lFjJ+X/5lMn6eBLCLx9O/QaJQiYXK5ISLn2nNzJFjUYfzTFDI5mJXkpczGSZO7uWCb/RaItYocDc4FZx\n",
    "mWCJCgvCz/YejeRygopswpSq0hJ+oJQxStQy0bga3Ej+6FxuUUZM2ol1hO8+zDkKrGRicGPky8sqtskD95iEDXxyOn3JNjln//UW\n",
    "Ei4kkxHusERhAo1gjZXWrfBI/SlKmZwOxEaEK11a1DLeed7f17hNUqbU87GmPZLMP64LjrkuiExfK1zITDbwFXUfB4AaFZYG9zor\n",
    "FJKhkXG7Bm/sdF/iULXJ08ph7XNSW21Pla3tVywlb6uWkT4rZxobk029fUq40n9AJnKiHw1m/mcrcOfNy+1PuLAduWPCJUrJfAHL\n",
    "52i3y6wOPIq58xfQUEqA\n",
    "\"\"\",\n",
    "    \"customers_raw.csv\": \"\"\"\n",
    "eNrFXEtvHbcV3udXzA+Ye8HD15C7BEIWRdOii6LoLnAVITAaO0HjFO0ucZcNEEmWYkUvS7IFBWhsKUmNOrbl/AXef1SQM3NnyHPI\n",
    "uc2rW0GX5JDn8Z3vfOT6Rx/eef/Wxl/evvlOffvGrY16/eadv9cf3nz39kcfvP3OjTsb9Ycb797auH2n3rh14+Z79Z/ev/3Rh6+t\n",
    "MQa1O3VPFx+7b91T913lnsxrt+ueu69rzriaMT0DXrvv3avFJ4u77tXiM3ftrhf/dN/V6+/d3Lh9hzF4/b0bf/3zjfnG327c+uC9\n",
    "jRo48wPzuopH3p5XtXvqnrurxd3w92fu2n29+If/j24ymIGs3bV75b7xU1TLOXgyB68kC7OI2h26K/d88Yl75l5W7jRZvpqBHY24\n",
    "HFAkA6owmqzdprt2/1583C55Z167l/7T3XP3jbvyY4oZszNQtTtyx27b/ct9sRxTphth2zWGf/YffeVeLD5dfFa53W6VmZ2QM2Zm\n",
    "XJLbXtW1H1PXbs8vyV23o+FV8hnX4y+vpf9GznmyY5fDjsk5M3M/f+nAOU8PQ7ab1/jNe+FPePFJu30P5t64Nt2R23dnM/fA7boz\n",
    "t+tO3JZ76E7cvW6lcgaCPPUmmcjPYmp37LfRfeuuKnffzzD3Qz/ttjI2KDEDRn+Mn/Dp4q4f0tZuPyz52j3zp9PuyL7b7Fa7547G\n",
    "K5YzYDPvNyWnsMnS/Q75pey7U7frjtyeu6zcMT44OQMI5lUYHFgyuIBgaAB1lR7BqXe6I7+1YZ5n7mv3bPGxn3EwZpHZIeNHBcDm\n",
    "69f93F25/3j7W3zamayY2BQAlax7eQTKjMx5L1jNA/d1zficKW+QorgfypARyMetB+7KfeOeuheVO5x7w482W3kv8e6J4gNwekgR\n",
    "DenDA723TM2BTa8cRCau+TC44166Z4u7ITx811olMhamZrxsiSApS1Tewg+CJW5W7qgQjvwZ8C4ojIKJqDqb01UbM6+9Oz7EawxO\n",
    "OLVGnduGpnYn7pV72VrGo3FwDyHOh4Oyd4OpEve+F+zryl2754u7+dAhZ352bBmGjvI+zZz5HfQu0u7GLra4ENmZpsalIgZnSbC+\n",
    "mFo7NJ3LZGL42lu/evO3v2ecvf7WG3/49RvzN//4xm9+99ab7Www8sJL8ijlzEdLMi1138ExHAjpweed5eAHyUGqmYDavXLf+0H7\n",
    "LFCLsLHehY5DUPs2xPsqF85gzmTB5fovF8mXd7NI78rP3HPvcv7wzpZRqI/KLCxktMTlJ6PEX/lvXmdcJaZ3nmwq6Dnwwmn1xuUz\n",
    "+XIPPvd7MLYrxucA3Zcju+KasqumTrFZXbn9OKSLGfCQPl+RH92QKIAjd7sg4xbMeBlW8tTPTLsT1lvOizD8VeU2SReDEBdGgKL7\n",
    "saDc6bqQHOWMybK5C0abu7fnaKr9dBtkOUN05iqAcFTBo024n2wCm8Bx3VEJUWO0UMdW0JQXOYRZIZPvPZ6MVHaw2oIdCEnagVBo\n",
    "8dthG47dqdt3226zBtEtPspc/b6qNAz4fLbOhEbjXmK008JrQxQBInW3NryIxke/FkpurrA5PTiZSJyioZGgMCFMRqBtxyPBNqL5\n",
    "sIMDRvdj4M2Q0Vs0Nv76IsJDQaE1NBth9sn8y9ScZeOZoPKkZNEMe+P0EpCpKO+jZOSpyXFCfDKvF58My/6eOjOZBZQScthXcmRx\n",
    "mxmoCrL8FZzYffARMypq769gfCLglMymtacqRRLoNxE64yqX1gKalnK0vcfZUCzYnOnpQCHpLLzGpIqM45xMGTILJfvhFS5GpU62\n",
    "4JAsjGzMavQhSOokBA0m0YzS/SERkzl23vClpg7l5Ym753bdF20B3h0JdDBepPm8ywbSRqUNgV2LxWK/S5ZOh4pNpW0xY01A3KVC\n",
    "D7lpV4UoiIZ/MmA36OAw5ZMKcsPFleN5DmTojHm30UOJ2u111dRVe45LQNnuJVCLEmQoUjJBbJepWfTcQTlGKJmtwRUqOY/iOOpp\n",
    "D00RNUoh12txq9LJqveyfA3IAb7jTdHY+VQTEIdfq3e9rWz8oMZrSFDxc9AQ4D0mqix8FVugQUM5xEzHLET8CWIL2ipZWb/Sb8KW\n",
    "vgiB/ap1/TEaWgLa4Of5j7DEzqwzzZBxPF4Nw+gZUOWuZkQa15DE07MkngJ0wKVs5BqowTlB7h7GiDxvgZrKrUwLYuf3l14DzWrL\n",
    "FSTZpGVk4Efpci0Fb/shJRlFtKrdE/di4EEiC7Fd2ZCrc3WWv9OaQP+46BOhhi4XErpBdONhOpga6psjtEiCNNYmOfrTZC9BdxCj\n",
    "TLxqQxpB8KhNt+0ptfb4EwyghnIhXS2FZhuGqKQnyRYQR94w0oqatMFzP91NGKIbHpTypCZOj6MsoebM0iCjZ/SamEH9coUgIj0P\n",
    "QXJmjaBWh6lTEpYBzDjVGmokvY8K4aMRPcSaELJJpqRRGZjR6CQzHIecvunuh57JVx0JA6EDFw/dQcgmhZCBL1pnTSjhjtpi80sS\n",
    "lKqpLkPTZJJ6k3rTg3QCUWa1+gkMHU7WWZN2ZHYmyQQ+Z2KFKW3mKAyjmlcVBqkeA1FY0jCSDjMxML1HduqaUOYXkrIBMpobnrj2\n",
    "Fhl1eVJ+9IOSGc3E/nlCAXVgM8EQXdsPS3ZWDXbKs0nXh8GM8OppWsioukonOhiIj57V5LpIuhmSHFpjBpNDx2RtoGJsP2yOppsH\n",
    "pplA996+V0AShkh+xiRR5mAFIiOEnQlq0pg0+LSst7EJGHiYxEqR3f/+M2ymH2RZBIm28/03O3AGtJ3aDHFrYVR7b5JNEDXLMqud\n",
    "odgoQ7YVJArszAQrKeyDJV3UppqHzdRYWAHIWkEEqnVmJcIcOz4AFlUKvilSrtgt7aZWRaVx6kT+9GTO8y2dTteZ1QQY3534iKWc\n",
    "A9EzNs2t3dIbFMkuViPUMpWQbTI5yZqRJd5blbTjVNPSmpxDWRTRdpZwTpB21G+Ppbx/DRgb1dAPl2P5nkQhQfduCYzRLB60QqXk\n",
    "dBG0sUOzmsJigMVKATAB41Gj+YIkV/jM/xsqy5E4KWRnYFRt+LikVwlB109xjaagKsR1YGnThWbtWiYXD0r6Jnh4Ew26XeIrIE/V\n",
    "A1MkIeKhfBJqLlYQCgUuwX/I0HPpV9z4Ab1a5yt32WLerXlB9SJ7qhGvuKG3xCAvOV+pcPH6I5EzRZPB1z9Pa2sNPPOTRK4dstEg\n",
    "Q0CkCg5glgqKAKyakBGKmdcyoA0HCjavg8+LiYmcpK3zAjoEAAqNA/AIPuxOdH9FMJK0xQdIG/RLsHzrADIIKpOAci+A23mpdAZV\n",
    "Ki4ASKrol5B9AYxlDadlTWTbAJuYKaMjAmiQPe2T0E7HUsTlwA115GO2OGWUWDMAdmxBhhrOkvz7dQ7f4gCwTKJIS9SFFvA+GLnp\n",
    "w5VqgQ4ilfI2Z2Tk5FCFKnrpc6ELfV2IzGJKYQSc4KSAp7XwQdn/JMZ9/fCkvBW4GOnc9ghuoC1sTLk/LsgOrUQqtxWIdRGIiOOE\n",
    "ZmvtnaskGn2ekQhPuC5XpL4AfP02wu/7kyAbWAC/ZCkGSKJkkA4hdGP2f5QMAfgYTx9kVEnAcUYFjirdDpT+TL11EMzLscYM2OcR\n",
    "iREQDM8QjpDRJIGAdNQv21En1wsd8EonAto8BK8S+9umaJIMlwaCZ4ZNFAetqxyt2PQDQbDF4HVLY2Ud4SWqsNOSGlKlNx9251X4\n",
    "0+Q+07cVQCgyJAkdwZnzjPigzBkBUiy1tYWgQXVJbN9aCe3jSKnU26SJ+MaLsvQ+dDDLmigQhsAz6yASVx0Xur0EBGxWrxIWK3EX\n",
    "9NFqdYAKlBdpQ0h71MVwGVPH22V41Nb+5Z1BGqSAtb3+aBTLzxMVjw4guIAfJZ0vPTk9Ut0kil/gU6nS0MNKUVdDC/PIx6+oKjAD\n",
    "zYDcSAoqz/DuplICrJ+sGBpV9kPa5MEZT5JH+x2Yb7ucKEemxDeAFFA9/JVx7+o0T5s2g0popC3yQ+i6Shd8Pq9WksTRjTaQhKjj\n",
    "l1Pogmzqamz4W6ktcTF0M/MyH/CiqyQobM1XNJ3c5aR+h6g6Qdq6ijuRFa3U5015cEs6mFdqjUbfp5GuCSgdHSkSaBFNsMBlJmPC\n",
    "nDXdTv/PbTBQMJnCDaFj73ov4MVeyfltzldI10t9LoJFiiIIghpsfP9hYvwujlH4CEnEuq7vT3XdAJREDNRJkixhKn5nhWagFKk7\n",
    "x2hchzkeZEMrKEW1v9dAjW9H7k8mTk1UUv0MmgreCl91fDxF9/gKiEpKqqHd0NDKoviWBtguRJUhHqFMazWfTZUITZ8TV0DijkJ7\n",
    "f2TK0pbck56wtFh1mhfx+Qt1fFKDB6R8bQ00i+8QRWv1Sc9OD60zOE2nApvzMsffuQ6JBZFwzY8f47MDMsCJ0IjDTJCmQpEWFSIc\n",
    "cOHA6RCvSX0qaBwy0LUUMejLUImtZUYMDTrlMg4TalnOyhddwKvTRkj6c1LYwWCKKNWaZtU4j+uww5Ss76lANc4V/b6Npd5Habxm\n",
    "q1zmA93QhL+XvSWQbSsRITcBPuFDNoQdWkQIP8ooi8ishaRuvdwHGpawKOeZJM40QaY2FOYAJHtLCVreQ90JYrMBcng+4qa2p7pX\n",
    "y+ZnDKt7FqkRhDyRBugCVrlXDEgY1/WUoZm65simYCOSx7WfoMLV8+gMKyqny5Bryd5Sk2bz1nub9JLFg+WSQZRlmyHaNaFhf+j2\n",
    "3FkgFL7IsFEds4DsFuvg2tTVxC8RHE5jxmZAuXgWQ26rHZnZGanF4lP8TpMT0/z09+PBxNc8TlP5djN0i5EfIwXda+vgK4dR7tvL\n",
    "XS+ndhRVC301nGEm8sjth3IVhlIA7KyseKWCqKE4TJPeDblAUX6CGTJ0DzCo6SLfu78ik9m1HUtTZruCRoduwtKKHqFIoqknCfpx\n",
    "M1kaWo1dchoXqRAcBiyIh6bVAcbgplrsSXqVm65gqGrfi+nS3lCRgqU0RoAEde35Wsxnnk8IJwwtnLBkEraUZucAa2taWjeEIw+b\n",
    "ztxxoJyPxuHBQkYYZOPL0CeFFwI6eMLKVaXlSNkU5vGoeck8FgUmrXaA6iRZ8vUBsDIhB7Bave2VFMzHUr0IL7SLbJO6SyGmbdNm\n",
    "GoFWI1D4OLP8iRhqNV1eeb0dep3nJXU5dsJ+MnJ2iGR2pxTyLOTYLgRYW5HN+/jKRqG1wBkbZfqtzO2ZchTnjLhTxdn4DvN94pmH\n",
    "lmjRpdKHM/pRj/8Ldc6ZiPPDAz8dWYaY8nYJ0qY5w9T8ffIyXf4GaD+DzMygEsD/eLUukgn3p9KgwpnCb0RwpkcGdYzOHQZ5yrgi\n",
    "4eGnIVNuhzLmINQ1m/1NnaH/bMKd+vEbG+GnpooP56KoTxIYdbfDYFnqo5RN4HmGlLMM5uWQlppbpGISkV4t5ucwppp30kTGO3xP\n",
    "wBIOJHfNfdDKFDfLuKmp4ShlOAcRct+ovDtZ0RV56H1SnBQHSjC+xokXqnZzavlil4NTL1Rxjx4TJzzMXJ5h8VH1OJJ36rIe56dX\n",
    "Er3IvKdljskP1xmEymGsTLlIGqYJFbccjS4juVeRoTuvnWa6GS4x4PHI28YcbArZH3oT6Bc4cB6FaA8kico5fmQD14tT7/pwJBTr\n",
    "XZMDGXbLkkAx8WQA57l3Nzj3IhX0EuLTH/oqY/FxRM7xWztHKz9DFgHJDhdxPhauPCapgVZMnnS9hu9XUWXxJEu9exbSTEJEjrRi\n",
    "y4NF9GhFGk5cWHQBEr1UtVVAaCgz8oaC8pwb/FBi4KuiBKOHcJ4jlzi3CZ92mXnzjfRhTtz+5YKhNwYPic7KBLgR5G1gLgDh9ewd\n",
    "J0OsWGRIFS54nPfvEfRfG62LVx254NSWiOi5woNV2rOt9InQb3NBJzSvBIsC5wm+tsZL79lwQrj/OGet5aiFBF9h+Dif7U4+ozRC\n",
    "SFRh1NUuP5Jf5SJ+h+MyIxKf+uKGxEdeHpYEqSGVQYm+HwKdQCnxMc1NtxdlS8ZJ+atkVA2EwsnyubPyFJK+//RfnulYMQ==\n",
    "\"\"\",\n",
    "    \"returns_raw.csv\": \"\"\"\n",
    "eNqtml+SHDUSxt85RR9A3a1P/3WLjr0A4cB+4MXeHcYP+2bwARgIvEHAeg3hA+ysgwEHXuMrqG+08aW6q1RD9yDt8GJ3uLJSysyf\n",
    "MlNZvnh0+fTi8cefPlRPLh4+uuCPv148efj0k0v+vKhPHz64fKT+9vTB48tPL/+uLh49+OzJ44/+ojUUNP/cISnEjcbGaOOVUeU/\n",
    "+2fluvxKIUMhaLVDUHy81lgjK6jyofxW3uyflZvyc/mp3JT35Xq1f14+8E0+Kr+Wa2qwosGonYYyaJf5UG7k/Z/2z8t/y3V5R3En\n",
    "4knttJ0X9Mqr8r7crGTV97K9N+Vtebf/fP/limvtn+8/5+uerxundjop46bV0Pl64Os2qB1/+fb1D/tn5W15U27KL+VdebsSffsv\n",
    "ym/7L+Sv8qZc759RR6QOp9VOZwXT6nh1yuBE8VDFJ4PDiIczNcSlBoMBDdDUkGz12lFDpIYZBQgvmYE0sxDt4kPhJKdGg1lrt4QJ\n",
    "RAHIaqfjLISqwTJyzpkm8O5A2im3Be4mJN+IhzVwa8tuWtBLKMwczhM6wS3A6IUR0DTiVXlRfigvyjfldflnuSrfU5ywwKNxiGHk\n",
    "ljaTBgS71Jl6gaSkRgiyBvRkgel8nWgY7dWOZhxWN7L6V9MWDaNvhECvdNhoe+cJJVnaBDrV3dLZsyVDO0xiRsHhdbvWsca5i1Zj\n",
    "RUNueLdrPZKTjDtqaPYArewZi8mF1U7t4GZxP7IgUbHGSl45aqAd/RqiaMjCO/xW2+2dQSI31tmadY8LunMoG3Jio5eYWj0hUE+m\n",
    "lodx4W/+YwuRJRU2BSFNp6122+HMaYmGAytSnFKATmM6CIcTT0el00a75sh3edoSDmd5ZoyUENfk73+Vr8V5/yjflBflZblaldfl\n",
    "Vbkq38mj78rVeCqzhMP50Ij7mhYHzI6iIy6WNKni8cdmr6iCyLhQ7cbsOTOyDXLk4tJyk5UlR44cuRQrInaj/ezY3gUcREf6/QID\n",
    "OgQzqVV58jdwrio4O4mbZXh6a4ATonIWrrVvDT+5IPHx0pj52UZ3VjyIuKnJLLcnrycfZ3o0O9TkljY6jbc8jvR51MiaxaHr2YIj\n",
    "ed7Uc6/N0j9dZ9aRO2/pgyynftYwh8GTP+/S4pz197I8I57O8r7yjbzVfju4U29EQ5pakrtZ8ITPS2IOcih9k5g9ufLZHdze7qaz\n",
    "lHp/1NAchZraX3YEzhO9IIGbj8ZY1vFRVNilCncuV/ok4m3zEabaf0KcXAQJOZTFIuWcKIOBhIRgxKFWtw7t4birJw0EIOrDfchv\n",
    "dJjSbGd5CWQimigIHZe5fXELToTSQogu7jSFYERbL4fTGmM1KZCOKIclKo2Njk1f2YVniKLBNV1MOODZZUSS18PidTPSKAbyEwNE\n",
    "A/NKbI5fJC0xuZr74xTJMyxGiHhu6IhrfStHRaEj60XYqsUndRKFJNfzmo/ittmfk4e2OSuROHYiEIlAMrWu8C4ch64ekdFPVi9X\n",
    "TwNtVIyiwVaHLfnpZTASgST1OU+7MDjXNkfGO4XUEBMPJaLH5kQiUqzpFLnB5WS6SRDx2ojw4pVG614iLomXKeFzq9P293UvCSTs\n",
    "u4xWJkyQLPJFEliyXlhu9cBhTcJLjs3tPq3hRqwhMlm3jVliM2X7NUTRkKWZwaKZ6ayIicBktjITtmmNeDaEWcSzHEAyOhzCTGay\n",
    "HDPXajhF6OpP6tcyscneN6HKTEUjJyuTqhxcvdK6dt9dbaeT12PTOKe1xUCkMnnLMrQyCm6j853pNwtc2TcL5kP67Y1UVOBkT9DQ\n",
    "ywW7vZao4zifPOwC+lw+ypniLMMTjfn8VQVai7hfiHPgckYcFHfN5QV6rU1/igaF2fnP86VJwyl4wa6ZjfoCPJP7sww4HwY7rXo7\n",
    "3+q8vSPqYIChZfyW5bTkwfMJ9t3QkmKt0nEDXTXYLs6hBZrs5skPPZTO9LfgEwC1pwC20CMdKFh2AFOnjQjTZn23uRwFA8fm0rUb\n",
    "eFlerASMl+WHclW+Lt+W78uP5d8rweF1+ZGvEynYwx3LThs4zTc4aQCEQKswi7N/AVsagP0jf8V2K73GONHQtFzQ65EOFJzyAcEt\n",
    "AD+644/jwfEAEA6DRtPY170B8oPjnQKtQ3uAgPAkRTnd2sCJcsZROcVvOSws+gkOyAHDMo1G6NwdDLw1wNgaY203mD8A9biQ6QvG\n",
    "xnoDnl/vdyHnrDDMAcdpL8BOdKDagTdSGLZqyJMOOoKgcjcwsRlI4/BdoytG9C9Mam45OH6/6XJQlNd9szpqx8KHSR7Gud5Qdxrx\n",
    "HpEwOSzUG63cgPd4gYHVLVbydajTQWwLYNFMFXH8PHVy9n3fYTs4iIZF05xwQTeQNziGhkX9HqbTFmYoj3MGDRuaOjk5rHcDXjTU\n",
    "wMO1G+htd+7bMN03eXIuDifNvJZaYpriwIE3nIwc5zN36NO7seThcCY0h8Pckcg43Ibj6G7KI2asmnACDic9c1xadLpd4rQbLjbt\n",
    "H8zQpQycdcNF2+Tq4+fHUyZy1g0nN9Igk2pJtqG32eGgGzKdb1YbOjecfcNLbTGzhjDSY4MDcXj56puUyZOXz/QgHF7D+0OBiZPN\n",
    "/UNMcIANH2rG0W6p4sSKWcSnlBr4JX3s+gUOsyGD4QpjOPw/iSEduOfllc3ZCF4yzf6/56uQefd9RqeQeXf/ZRsy4L7HhP1/mlrO\n",
    "AQ==\n",
    "\"\"\",\n",
    "}\n",
    "os.makedirs(\"data\", exist_ok=True)\n",
    "for _name, _packed in _DATA.items():\n",
    "    with open(os.path.join(\"data\", _name), \"wb\") as _f:\n",
    "        _f.write(zlib.decompress(base64.b64decode(_packed)))\n",
    "print(\"Файлы данных:\", \", \".join(sorted(_DATA)))\n",
    "\n",
    "# ─── Проверка упражнений: тот же раннер, что у задач на сайте ───\n",
    "# ─── Запуск тестов ───────────────────────────────────────────────────────────\n",
    "\n",
    "_TIMEOUT = 2.0  # секунд на тест; отдельный тест может задать свой: test_x.timeout = 10\n",
    "\n",
    "\n",
    "def _call_with_timeout(target, seconds: float) -> bool:\n",
    "    \"\"\"Запускает target() в отдельном потоке. Возвращает False, если за seconds он не завершился.\"\"\"\n",
    "    import sys\n",
    "    import threading\n",
    "\n",
    "    if sys.platform == \"emscripten\":  # Python в браузере: потоков нет, таймаут обеспечивает страница\n",
    "        target()\n",
    "        return True\n",
    "    # Большой стек нужен, чтобы глубокая рекурсия давала RecursionError, а не аварийное завершение Python.\n",
    "    for megabytes in (256, 64, 0):\n",
    "        try:\n",
    "            threading.stack_size(megabytes * 1024 * 1024)\n",
    "            worker = threading.Thread(target=target, daemon=True)\n",
    "            worker.start()\n",
    "        except (ValueError, RuntimeError, MemoryError):\n",
    "            continue\n",
    "        worker.join(seconds)\n",
    "        return not worker.is_alive()\n",
    "    target()  # потоки недоступны — запускаем без таймаута\n",
    "    return True\n",
    "\n",
    "\n",
    "def _test_info(fn) -> dict:\n",
    "    \"\"\"Описание теста, его лимит времени и сообщение на случай зависания.\"\"\"\n",
    "    limit = getattr(fn, \"timeout\", _TIMEOUT)\n",
    "    hint = getattr(fn, \"timeout_hint\", \"\")\n",
    "    message = (\n",
    "        f\"тест не завершился за {limit:g} с — возможно, бесконечный цикл или рекурсия: \"\n",
    "        \"проверьте, что границы/аргументы изменяются на каждом шаге\" + (f\". {hint}\" if hint else \"\")\n",
    "    )\n",
    "    return {\"title\": (fn.__doc__ or fn.__name__).strip(), \"timeout\": limit, \"timeout_message\": message}\n",
    "\n",
    "\n",
    "def _error_result(title: str, error: BaseException, user_file: str | None) -> dict:\n",
    "    \"\"\"Результат упавшего теста: статус, сообщение, тип исключения и строка в коде решения.\"\"\"\n",
    "    line = None\n",
    "    if user_file:  # самый глубокий кадр в коде решения; ошибка внутри самих тестов — без номера строки\n",
    "        tb = error.__traceback__\n",
    "        while tb is not None:\n",
    "            if tb.tb_frame.f_code.co_filename == user_file:\n",
    "                line = tb.tb_lineno\n",
    "            tb = tb.tb_next\n",
    "    status, message = \"error\", f\"{type(error).__name__}: {error}\"\n",
    "    if isinstance(error, NotImplementedError):\n",
    "        status, message, line = \"not_written\", \"функция ещё не написана\", None\n",
    "    elif isinstance(error, AssertionError):\n",
    "        status, message = \"failed\", str(error) or \"проверка не прошла\"\n",
    "    elif isinstance(error, RecursionError):\n",
    "        message = \"RecursionError: слишком глубокая рекурсия — проверьте базовый случай и то, что задача уменьшается\"\n",
    "    elif isinstance(error, MemoryError) or (isinstance(error, OverflowError) and \"index-sized\" in str(error)):\n",
    "        message = f\"{type(error).__name__}: не хватило памяти — возможно, создаётся слишком большая структура данных\"\n",
    "    return {\"title\": title, \"status\": status, \"message\": message, \"error_type\": type(error).__name__, \"line\": line}\n",
    "\n",
    "\n",
    "def _run_tests(verbose: bool = True, on_start=None, on_result=None, user_file: str | None = None) -> list[dict]:\n",
    "    \"\"\"Запускает тесты задачи (_TESTS — функции из tests.py) по порядку, каждый — с таймаутом.\n",
    "\n",
    "    Возвращает список результатов: {\"title\", \"status\", \"message\", \"error_type\", \"line\"}, где status —\n",
    "    passed · failed · error · not_written · timeout · not_run. Если тест завис, при verbose=True печатает\n",
    "    итог и завершает процесс: зависший поток остановить нельзя. on_start(i) и on_result(i, результат)\n",
    "    вызываются по мере выполнения; user_file — имя файла с кодом решения для номеров строк в ошибках.\n",
    "    \"\"\"\n",
    "    import os\n",
    "    import sys\n",
    "\n",
    "    if verbose:\n",
    "        try:  # ✓ и кириллица в консоли Windows со старой кодировкой\n",
    "            sys.stdout.reconfigure(encoding=\"utf-8\")\n",
    "        except Exception:  # noqa: BLE001 — reconfigure может не быть (перенаправленный вывод, браузер)\n",
    "            pass\n",
    "\n",
    "    tests = list(_TESTS)\n",
    "    results: list[dict] = []\n",
    "    not_written = hung = False\n",
    "    for index, fn in enumerate(tests):\n",
    "        info = _test_info(fn)\n",
    "        outcome: dict = {}\n",
    "\n",
    "        def target(fn=fn, outcome=outcome):\n",
    "            try:\n",
    "                fn()\n",
    "            except BaseException as e:\n",
    "                outcome[\"error\"] = e\n",
    "\n",
    "        if on_start:\n",
    "            on_start(index)\n",
    "        if not _call_with_timeout(target, info[\"timeout\"]):\n",
    "            results.append({\"title\": info[\"title\"], \"status\": \"timeout\", \"message\": info[\"timeout_message\"], \"error_type\": None, \"line\": None})\n",
    "            for other in tests[index + 1:]:\n",
    "                results.append({\"title\": _test_info(other)[\"title\"], \"status\": \"not_run\", \"message\": \"не запускался: предыдущий тест завис\", \"error_type\": None, \"line\": None})\n",
    "            hung = True\n",
    "            break\n",
    "        error = outcome.get(\"error\")\n",
    "        if error is None:\n",
    "            result = {\"title\": info[\"title\"], \"status\": \"passed\", \"message\": \"\", \"error_type\": None, \"line\": None}\n",
    "        else:\n",
    "            result = _error_result(info[\"title\"], error, user_file)\n",
    "            not_written = not_written or result[\"status\"] == \"not_written\"\n",
    "        results.append(result)\n",
    "        if on_result:\n",
    "            on_result(index, result)\n",
    "\n",
    "    if not_written:  # пока функция не написана, «даром» пройденные тесты не засчитываются\n",
    "        for r in results:\n",
    "            if r[\"status\"] == \"passed\":\n",
    "                r.update(status=\"not_written\", message=\"не засчитан, пока функция не написана\")\n",
    "    for r in results:\n",
    "        if len(r[\"message\"]) > 300:\n",
    "            r[\"message\"] = r[\"message\"][:300] + \"…\"\n",
    "\n",
    "    if verbose:\n",
    "        if not_written:\n",
    "            print(\"Функция ещё не написана — замените raise NotImplementedError своим кодом.\\n\")\n",
    "        for r in results:\n",
    "            mark = {\"passed\": \"✓\", \"not_run\": \"–\"}.get(r[\"status\"], \"✗\")\n",
    "            print(f\"{mark} {r['title']}\")\n",
    "            if r[\"message\"]:\n",
    "                print(f\"    {r['message']}\")\n",
    "        passed = sum(r[\"status\"] == \"passed\" for r in results)\n",
    "        print(\"─\" * 40)\n",
    "        print(f\"Прошло {passed} из {len(results)}\" + (\" — всё верно!\" if passed == len(results) else \"\"))\n",
    "        if hung:\n",
    "            print(\"Выполнение остановлено: один из тестов завис.\")\n",
    "            sys.stdout.flush()\n",
    "            os._exit(1)\n",
    "    return results\n",
    "\n",
    "\n",
    "def _check(*tests, names=()):\n",
    "    \"\"\"Запускает проверки упражнения и печатает ✓ / ✗ и итог — как на сайте.\"\"\"\n",
    "    for name in names:\n",
    "        if \".\" in name:  # «Класс.метод» из заготовки: класс должен быть объявлен\n",
    "            if not isinstance(globals().get(name.split(\".\")[0]), type):\n",
    "                print(f\"✗ класс {name.split('.')[0]} не объявлен — выполните ячейку упражнения\")\n",
    "                return\n",
    "            continue\n",
    "        if name not in globals():\n",
    "            print(f\"✗ переменная {name} не создана — выполните ячейку упражнения\")\n",
    "            return\n",
    "        if globals()[name] is Ellipsis:\n",
    "            print(f\"✗ {name} пока равна ... — замените многоточие своим кодом\")\n",
    "            return\n",
    "    globals()[\"_TESTS\"] = list(tests)\n",
    "    _run_tests()"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "setup"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Вторая часть итогового проекта. Таблица собрана — теперь нужно ответить на вопросы владельца: как шёл год,\n",
    "на чём магазин зарабатывает, где и кому продаёт. В отчёте везде используется **чистая** выручка `net` —\n",
    "за вычетом возвратов.\n",
    "\n",
    "Первая ячейка заново собирает таблицу из прошлого урока — то же самое, записанное тремя цепочками, плюс один\n",
    "новый столбец: `month`, номер месяца. Прочитайте её: каждая строка вам знакома."
   ],
   "id": "text-11"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Рабочая таблица: 2448 строк, 17 столбцов\n",
    "import pandas as pd\n",
    "\n",
    "short = {\"Спб\": \"Санкт-Петербург\", \"С.-Петербург\": \"Санкт-Петербург\", \"Екб\": \"Екатеринбург\"}\n",
    "customers = (\n",
    "    pd.read_csv(\"data/customers_raw.csv\")\n",
    "    .assign(\n",
    "        customer_id=lambda d: d[\"customer_id\"].str.strip().str.upper(),\n",
    "        city=lambda d: d[\"city\"].str.strip().str.title().replace(short),\n",
    "        segment=lambda d: d[\"segment\"].str.strip().str.lower(),\n",
    "    )\n",
    "    [[\"customer_id\", \"city\", \"segment\"]]\n",
    "    .drop_duplicates()\n",
    ")\n",
    "returns = (\n",
    "    pd.read_csv(\"data/returns_raw.csv\")\n",
    "    .drop_duplicates()\n",
    "    .drop_duplicates(subset=[\"order_id\", \"product_id\"])\n",
    "    [[\"order_id\", \"product_id\", \"quantity\"]]\n",
    "    .rename(columns={\"quantity\": \"returned\"})\n",
    ")\n",
    "data = (\n",
    "    pd.read_csv(\"data/orders.csv\", parse_dates=[\"date\"])\n",
    "    .merge(pd.read_csv(\"data/products.csv\"), on=\"product_id\", validate=\"many_to_one\")\n",
    "    .merge(customers, on=\"customer_id\", validate=\"many_to_one\")\n",
    "    .merge(returns, on=[\"order_id\", \"product_id\"], how=\"left\", validate=\"one_to_one\")\n",
    "    .assign(\n",
    "        returned=lambda d: d[\"returned\"].fillna(0).astype(\"int64\"),\n",
    "        revenue=lambda d: d[\"price\"] * d[\"quantity\"],\n",
    "        net=lambda d: d[\"price\"] * (d[\"quantity\"] - d[\"returned\"]),\n",
    "        profit=lambda d: (d[\"price\"] - d[\"cost\"]) * (d[\"quantity\"] - d[\"returned\"]),\n",
    "        month=lambda d: d[\"date\"].dt.month,\n",
    "    )\n",
    ")\n",
    "print(data.shape, data[\"net\"].sum(), data[\"profit\"].sum())"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "build"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Вопрос 1. Как шёл год"
   ],
   "id": "text-20"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 1. Динамика по месяцам\n",
    "\n",
    "Сохраните в `by_month` таблицу по месяцам — сгруппируйте `data` по столбцу `month`: `net` и `profit` — суммы, `growth` — рост `net` к прошлому месяцу\n",
    "(три знака). В `best_month` и `worst_month` сохраните номера лучшего и худшего месяца по чистой выручке.\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "`data.groupby(\"month\").agg(net=(\"net\", \"sum\"), profit=(\"profit\", \"sum\"))`.\n",
    "\n",
    "</details>\n",
    "<details><summary>Подсказка 2</summary>\n",
    "\n",
    "`by_month[\"net\"].pct_change().round(3)`; лучший и худший — `idxmax()` и `idxmin()`.\n",
    "\n",
    "</details>"
   ],
   "id": "months-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "by_month = ...\n",
    "# добавьте в by_month столбец growth\n",
    "best_month = ...\n",
    "worst_month = ..."
   ],
   "execution_count": null,
   "outputs": [],
   "id": "months"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 1 — выполните после своего решения\n",
    "def test_by_month():\n",
    "    \"by_month — чистая выручка, прибыль и рост по месяцам\"\n",
    "    assert isinstance(by_month, pd.DataFrame), f\"by_month — это {type(by_month).__name__}, а нужна таблица\"\n",
    "    assert list(by_month.index) == list(range(1, 13)), \"в индексе by_month должны быть номера месяцев 1–12: группируйте по столбцу month\"\n",
    "    assert list(by_month.columns) == [\"net\", \"profit\", \"growth\"], f\"столбцы сейчас {list(by_month.columns)}, а нужны net, profit, growth\"\n",
    "    assert len(by_month) == 12 and by_month.loc[1, \"net\"] == 297120 and by_month.loc[1, \"profit\"] == 140940, \"в by_month 12 месяцев, net и profit — суммы за месяц\"\n",
    "    assert by_month[\"growth\"].isna().sum() == 1, \"у января нет прошлого месяца — первое значение growth должно быть пропуском\"\n",
    "    assert abs(by_month.loc[2, \"growth\"] - (-0.179)) < 1e-9 and abs(by_month.loc[12, \"growth\"] - 0.296) < 1e-9, \"growth не тот: нужен относительный рост net к прошлому месяцу, три знака\"\n",
    "\n",
    "\n",
    "def test_best():\n",
    "    \"best_month и worst_month — лучший и худший месяц по чистой выручке\"\n",
    "    assert best_month == 12, f\"best_month = {best_month!r} — это не месяц с наибольшей чистой выручкой\"\n",
    "    assert worst_month == 7, f\"worst_month = {worst_month!r} — это не месяц с наименьшей чистой выручкой\"\n",
    "\n",
    "_check(test_by_month, test_best, names=[\"by_month\",\"best_month\",\"worst_month\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "months-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "by_month = data.groupby(\"month\").agg(net=(\"net\", \"sum\"), profit=(\"profit\", \"sum\"))\n",
    "by_month[\"growth\"] = by_month[\"net\"].pct_change().round(3)\n",
    "best_month = by_month[\"net\"].idxmax()\n",
    "worst_month = by_month[\"net\"].idxmin()\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "months-solution"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Месяцы\n",
    "by_month"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "months-view"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Вопрос 2. На чём зарабатываем"
   ],
   "id": "text-32"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 2. Категории: прибыль, маржа, возвраты\n",
    "\n",
    "Сохраните в `by_category` таблицу по категориям: `net`, `profit` — суммы; `sold` — продано штук (`quantity`);\n",
    "`returned` — возвращено штук; `margin` — прибыль, делённая на чистую выручку; `return_rate` — доля возвращённых\n",
    "штук. Доли — три знака. Отсортируйте по убыванию прибыли.\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "Четыре показателя — в одном `agg`; два вычисляемых столбца — после него.\n",
    "\n",
    "</details>\n",
    "<details><summary>Подсказка 2</summary>\n",
    "\n",
    "`(by_category[\"profit\"] / by_category[\"net\"]).round(3)` и `(by_category[\"returned\"] / by_category[\"sold\"]).round(3)`.\n",
    "\n",
    "</details>"
   ],
   "id": "categories-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "by_category = ..."
   ],
   "execution_count": null,
   "outputs": [],
   "id": "categories"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 2 — выполните после своего решения\n",
    "def test_by_category():\n",
    "    \"by_category — выручка, прибыль, штуки, маржа и доля возврата по категориям\"\n",
    "    assert isinstance(by_category, pd.DataFrame), f\"by_category — это {type(by_category).__name__}, а нужна таблица\"\n",
    "    assert list(by_category.columns) == [\"net\", \"profit\", \"sold\", \"returned\", \"margin\", \"return_rate\"], f\"столбцы сейчас {list(by_category.columns)}, а нужны net, profit, sold, returned, margin, return_rate\"\n",
    "    assert len(by_category) == 5 and by_category.loc[\"Кофе\", \"net\"] == 1830330 and by_category.loc[\"Кофе\", \"sold\"] == 2138, \"в by_category пять категорий, net и sold — суммы по категории\"\n",
    "    assert abs(by_category.loc[\"Чай\", \"margin\"] - 0.56) < 1e-9, \"margin не тот: нужна доля прибыли в net, три знака\"\n",
    "    assert abs(by_category.loc[\"Кофе\", \"return_rate\"] - 0.042) < 1e-9, \"return_rate не тот: нужна доля возвращённых штук среди проданных, три знака\"\n",
    "\n",
    "\n",
    "def test_order():\n",
    "    \"категории — по убыванию прибыли\"\n",
    "    assert isinstance(by_category, pd.DataFrame) and \"profit\" in by_category.columns, \"сначала исправьте то, о чём говорит проверка выше\"\n",
    "    assert by_category[\"profit\"].tolist() == sorted(by_category[\"profit\"].tolist(), reverse=True), \"отсортируйте by_category по убыванию profit\"\n",
    "\n",
    "_check(test_by_category, test_order, names=[\"by_category\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "categories-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "by_category = data.groupby(\"category\").agg(net=(\"net\", \"sum\"), profit=(\"profit\", \"sum\"), sold=(\"quantity\", \"sum\"), returned=(\"returned\", \"sum\"))\n",
    "by_category[\"margin\"] = (by_category[\"profit\"] / by_category[\"net\"]).round(3)\n",
    "by_category[\"return_rate\"] = (by_category[\"returned\"] / by_category[\"sold\"]).round(3)\n",
    "by_category = by_category.sort_values(\"profit\", ascending=False)\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "categories-solution"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Категории\n",
    "by_category"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "categories-view"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Вопрос 3. Где продаём"
   ],
   "id": "text-45"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 3. Города и средний чек\n",
    "\n",
    "Сохраните в `by_city` таблицу по городам: `net` — чистая выручка, `orders` — число разных заказов, `buyers` —\n",
    "число разных покупателей, `check` — средний чек (`net` на заказ, до рублей), `share` — доля города в общей\n",
    "чистой выручке (три знака). В `top_check_city` сохраните город с наибольшим средним чеком.\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "`orders=(\"order_id\", \"nunique\")`, `buyers=(\"customer_id\", \"nunique\")`.\n",
    "\n",
    "</details>\n",
    "<details><summary>Подсказка 2</summary>\n",
    "\n",
    "`(by_city[\"net\"] / by_city[\"orders\"]).round()` и `(by_city[\"net\"] / by_city[\"net\"].sum()).round(3)`.\n",
    "\n",
    "</details>"
   ],
   "id": "cities-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "by_city = ...\n",
    "top_check_city = ..."
   ],
   "execution_count": null,
   "outputs": [],
   "id": "cities"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 3 — выполните после своего решения\n",
    "def test_by_city():\n",
    "    \"by_city — выручка, заказы, покупатели, средний чек и доля по городам\"\n",
    "    assert isinstance(by_city, pd.DataFrame), f\"by_city — это {type(by_city).__name__}, а нужна таблица\"\n",
    "    assert list(by_city.columns) == [\"net\", \"orders\", \"buyers\", \"check\", \"share\"], f\"столбцы сейчас {list(by_city.columns)}, а нужны net, orders, buyers, check, share\"\n",
    "    assert by_city.loc[\"Москва\", \"net\"] == 1245060, \"net не тот: нужна сумма net по городу\"\n",
    "    assert by_city.loc[\"Москва\", \"orders\"] != 948, \"orders — число разных заказов, а не строк\"\n",
    "    assert by_city.loc[\"Москва\", \"orders\"] == 627 and by_city.loc[\"Москва\", \"buyers\"] == 79, \"orders или buyers не те: нужно число разных заказов и покупателей\"\n",
    "    assert by_city.loc[\"Москва\", \"check\"] == 1986, \"check не тот: нужна net на один заказ, до рублей\"\n",
    "    assert abs(by_city.loc[\"Москва\", \"share\"] - 0.394) < 1e-9 and abs(by_city[\"share\"].sum() - 1) < 0.01, \"share не тот: нужна доля города в общей net, три знака\"\n",
    "\n",
    "\n",
    "def test_top():\n",
    "    \"top_check_city — город с наибольшим средним чеком\"\n",
    "    assert top_check_city == \"Санкт-Петербург\", f\"top_check_city = {top_check_city!r}, а самый высокий средний чек в другом городе\"\n",
    "\n",
    "_check(test_by_city, test_top, names=[\"by_city\",\"top_check_city\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "cities-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "by_city = data.groupby(\"city\").agg(net=(\"net\", \"sum\"), orders=(\"order_id\", \"nunique\"), buyers=(\"customer_id\", \"nunique\"))\n",
    "by_city[\"check\"] = (by_city[\"net\"] / by_city[\"orders\"]).round()\n",
    "by_city[\"share\"] = (by_city[\"net\"] / by_city[\"net\"].sum()).round(3)\n",
    "top_check_city = by_city[\"check\"].idxmax()\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "cities-solution"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Вопрос 4. Что и где"
   ],
   "id": "text-56"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 4. Категории × города\n",
    "\n",
    "Сохраните в `cross` сводную таблицу чистой выручки: категории в строках, города в столбцах, с итогами «Всего».\n",
    "В `moscow_coffee_share` сохраните долю московского кофе во всей чистой выручке магазина — посчитайте её\n",
    "по ячейкам `cross`.\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "`data.pivot_table(values=\"net\", index=\"category\", columns=\"city\", aggfunc=\"sum\", margins=True, margins_name=\"Всего\")`.\n",
    "\n",
    "</details>\n",
    "<details><summary>Подсказка 2</summary>\n",
    "\n",
    "`cross.loc[\"Кофе\", \"Москва\"] / cross.loc[\"Всего\", \"Всего\"]`.\n",
    "\n",
    "</details>"
   ],
   "id": "pivot-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "cross = ...\n",
    "moscow_coffee_share = ..."
   ],
   "execution_count": null,
   "outputs": [],
   "id": "pivot"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 4 — выполните после своего решения\n",
    "def test_cross():\n",
    "    \"cross — чистая выручка: категории × города, с итогами «Всего»\"\n",
    "    assert isinstance(cross, pd.DataFrame), f\"cross — это {type(cross).__name__}, а нужна таблица\"\n",
    "    assert \"Всего\" in cross.index and \"Всего\" in cross.columns, \"в cross нет итогов: вспомните параметры margins и margins_name\"\n",
    "    assert cross.shape == (6, 6), f\"у cross размер {cross.shape}, а нужно 6 × 6: пять категорий и пять городов с итогами\"\n",
    "    assert \"Кофе\" in cross.index and \"Москва\" in cross.columns, \"в строках — категории, в столбцах — города\"\n",
    "    assert cross.loc[\"Кофе\", \"Москва\"] == 735480 and cross.loc[\"Всего\", \"Всего\"] == 3159140, \"в ячейках должна быть сумма net\"\n",
    "\n",
    "\n",
    "def test_share():\n",
    "    \"moscow_coffee_share — доля московского кофе во всей чистой выручке\"\n",
    "    assert abs(moscow_coffee_share - 735480 / 3159140) < 1e-9, f\"moscow_coffee_share = {moscow_coffee_share!r} — это не доля московского кофе во всей чистой выручке\"\n",
    "\n",
    "_check(test_cross, test_share, names=[\"cross\",\"moscow_coffee_share\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "pivot-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "cross = data.pivot_table(values=\"net\", index=\"category\", columns=\"city\", aggfunc=\"sum\", margins=True, margins_name=\"Всего\")\n",
    "moscow_coffee_share = cross.loc[\"Кофе\", \"Москва\"] / cross.loc[\"Всего\", \"Всего\"]\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "pivot-solution"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Сводная\n",
    "cross"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "pivot-view"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Вопрос 5. Кому продаём"
   ],
   "id": "text-69"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 5. Сегменты и лучшие покупатели\n",
    "\n",
    "Сохраните в `by_segment` таблицу по сегментам: `net`, `buyers` (число разных покупателей) и `per_buyer` —\n",
    "чистая выручка на покупателя, до рублей. В `vip` сохраните пять покупателей с наибольшей чистой выручкой:\n",
    "столбцы `customer_id`, `city`, `segment`, `net`, индекс с нуля.\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "Для `vip` — группировка по трём столбцам с `as_index=False`, сумма `net`, `.nlargest(5, \"net\")`.\n",
    "\n",
    "</details>"
   ],
   "id": "segments-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "by_segment = ...\n",
    "vip = ..."
   ],
   "execution_count": null,
   "outputs": [],
   "id": "segments"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 5 — выполните после своего решения\n",
    "def test_segment():\n",
    "    \"by_segment — выручка, покупатели и выручка на покупателя по сегментам\"\n",
    "    assert isinstance(by_segment, pd.DataFrame), f\"by_segment — это {type(by_segment).__name__}, а нужна таблица\"\n",
    "    assert list(by_segment.columns) == [\"net\", \"buyers\", \"per_buyer\"], f\"столбцы сейчас {list(by_segment.columns)}, а нужны net, buyers, per_buyer\"\n",
    "    assert by_segment.loc[\"оптовый\", \"net\"] == 737000 and by_segment.loc[\"оптовый\", \"buyers\"] == 25, \"значения не те: net — сумма по сегменту, buyers — число разных покупателей\"\n",
    "    assert by_segment.loc[\"оптовый\", \"per_buyer\"] == 29480, \"per_buyer не тот: нужна net на одного покупателя, до рублей\"\n",
    "\n",
    "\n",
    "def test_vip():\n",
    "    \"vip — пять покупателей с наибольшей чистой выручкой\"\n",
    "    assert isinstance(vip, pd.DataFrame), f\"vip — это {type(vip).__name__}, а нужна таблица\"\n",
    "    assert list(vip.columns) == [\"customer_id\", \"city\", \"segment\", \"net\"], f\"столбцы сейчас {list(vip.columns)}, а нужны customer_id, city, segment, net\"\n",
    "    assert len(vip) == 5, f\"в vip {len(vip)} строк, а нужно пять\"\n",
    "    assert vip[\"customer_id\"].tolist() == [\"C134\", \"C137\", \"C118\", \"C151\", \"C216\"], f\"покупатели сейчас {vip['customer_id'].tolist()}: нужны пять покупателей с наибольшей net\"\n",
    "    assert list(vip.index) == [0, 1, 2, 3, 4], \"индекс vip должен идти с нуля\"\n",
    "\n",
    "_check(test_segment, test_vip, names=[\"by_segment\",\"vip\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "segments-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "by_segment = data.groupby(\"segment\").agg(net=(\"net\", \"sum\"), buyers=(\"customer_id\", \"nunique\"))\n",
    "by_segment[\"per_buyer\"] = (by_segment[\"net\"] / by_segment[\"buyers\"]).round()\n",
    "vip = data.groupby([\"customer_id\", \"city\", \"segment\"], as_index=False)[\"net\"].sum().nlargest(5, \"net\").reset_index(drop=True)\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "segments-solution"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Графики"
   ],
   "id": "text-79"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Упражнение 6. Год на одном графике\n",
    "\n",
    "Постройте линейный график столбцов `net` и `profit` таблицы `by_month` с заголовком «Чистая выручка и прибыль\n",
    "по месяцам, 2025» и сохраните в `ax`. Подпишите оси: X — «месяц», Y — «₽».\n",
    "\n",
    "<details><summary>Подсказка 1</summary>\n",
    "\n",
    "`by_month[[\"net\", \"profit\"]].plot(title=\"...\")`.\n",
    "\n",
    "</details>"
   ],
   "id": "chart-task"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "ax = ...\n",
    "# подпишите оси"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "chart"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Проверка упражнения 6 — выполните после своего решения\n",
    "def test_chart():\n",
    "    \"ax — две линии: чистая выручка и прибыль по месяцам\"\n",
    "    assert hasattr(ax, \"lines\") and hasattr(ax, \"get_title\"), f\"ax — это {type(ax).__name__}, а нужен график\"\n",
    "    assert len(ax.lines) != 3, \"на графике три линии: столбец growth рисовать не нужно — выберите net и profit\"\n",
    "    assert len(ax.lines) == 2, f\"линий на графике: {len(ax.lines)}, а нужны две — net и profit\"\n",
    "    assert [line.get_label() for line in ax.lines] == [\"net\", \"profit\"], \"линии должны быть net и profit — в этом порядке\"\n",
    "    assert list(ax.lines[0].get_ydata())[0] == 297120 and len(ax.lines[0].get_ydata()) == 12, \"на графике должны быть 12 месяцев из by_month\"\n",
    "\n",
    "\n",
    "def test_labels():\n",
    "    \"заголовок и подписи осей\"\n",
    "    assert hasattr(ax, \"get_title\"), \"сначала постройте график и сохраните его в ax\"\n",
    "    assert ax.get_title() == \"Чистая выручка и прибыль по месяцам, 2025\", f\"заголовок сейчас {ax.get_title()!r}\"\n",
    "    assert ax.get_xlabel() == \"месяц\" and ax.get_ylabel() == \"₽\", f\"подписи осей сейчас {ax.get_xlabel()!r} и {ax.get_ylabel()!r}, а нужны «месяц» и «₽»\"\n",
    "\n",
    "_check(test_chart, test_labels, names=[\"ax\"])"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "chart-check"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<details><summary>Решение</summary>\n",
    "\n",
    "```python\n",
    "ax = by_month[[\"net\", \"profit\"]].plot(title=\"Чистая выручка и прибыль по месяцам, 2025\")\n",
    "ax.set_xlabel(\"месяц\")\n",
    "ax.set_ylabel(\"₽\")\n",
    "```\n",
    "\n",
    "</details>"
   ],
   "id": "chart-solution"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Прибыль по категориям\n",
    "by_category[\"profit\"].sort_values().plot(kind=\"barh\", title=\"Прибыль по категориям, ₽\")"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "category-chart"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Отчёт"
   ],
   "id": "text-90"
  },
  {
   "cell_type": "code",
   "metadata": {},
   "source": [
    "# Главные числа\n",
    "total_net = data[\"net\"].sum()\n",
    "total_profit = data[\"profit\"].sum()\n",
    "print(\"Выручка:\", data[\"revenue\"].sum(), \"₽ · после возвратов:\", total_net, \"₽\")\n",
    "print(\"Прибыль:\", total_profit, \"₽ · маржа:\", round(total_profit / total_net * 100, 1), \"%\")\n",
    "print(\"Заказов:\", data[\"order_id\"].nunique(), \"· покупателей:\", data[\"customer_id\"].nunique())\n",
    "print(\"Лучший месяц:\", best_month, \"· худший:\", worst_month)\n",
    "print(\"Самый высокий средний чек:\", top_check_city)\n",
    "print(by_segment)"
   ],
   "execution_count": null,
   "outputs": [],
   "id": "summary"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Выводы"
   ],
   "id": "text-94"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*После решения всех упражнений урока:*\n",
    "\n",
    "**Год в целом.** Магазин продал товаров на 3.30 млн ₽; после возвратов осталось 3.16 млн ₽, из них прибыль —\n",
    "1.48 млн ₽, маржа 46.7 %. 1576 заказов от 213 покупателей.\n",
    "\n",
    "**Сезонность.** Лучший месяц — декабрь (361 тыс. ₽, +30 % к ноябрю), худший — июль (171 тыс. ₽, −32 %\n",
    "к июню). Лето — провал, четвёртый квартал — пик. К летнему спаду стоит готовиться заранее: акциями или\n",
    "холодными напитками в ассортименте.\n",
    "\n",
    "**Категории.** Больше половины прибыли даёт кофе, но маржа у него самая низкая — 43 %. Самая маржинальная\n",
    "категория — чай (56 %), и он же хорошо продаётся зимой. Возвращают 4 % проданных штук, и доля возвратов\n",
    "по категориям почти одинакова — проблемной категории нет.\n",
    "\n",
    "**Города.** Москва — 39 % чистой выручки, Санкт-Петербург — четверть. Средний чек по городам отличается\n",
    "мало: от 1860 ₽ в Новосибирске до 2100 ₽ в Санкт-Петербурге.\n",
    "\n",
    "**Покупатели.** 25 оптовых покупателей приносят почти четверть выручки: на каждого приходится 29 тыс. ₽\n",
    "в год — в два с лишним раза больше, чем на обычного. Четверо из пяти лучших покупателей — оптовые. Потеря\n",
    "даже нескольких таких клиентов заметно ударит по выручке."
   ],
   "id": "component-97"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Что вы теперь умеете\n",
    "\n",
    "За курс вы прошли весь путь аналитика: загрузить данные, осмотреть, почистить, соединить, посчитать,\n",
    "свернуть, показать и объяснить. Этот отчёт — не учебное упражнение, а обычная рабочая задача, и вы решили её\n",
    "сами.\n",
    "\n",
    "Дальше — практика на своих данных. Возьмите любую выгрузку, которая вам интересна, и задайте ей вопросы.\n",
    "Когда забудется название метода или параметра — [справочник](https://ds-edu-trainer.vercel.app/reference/pandas/cheatsheet) под рукой."
   ],
   "id": "text-116"
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Подробнее в справочнике:** [pandas/groupby-agg](https://ds-edu-trainer.vercel.app/reference/pandas/groupby-agg) · [pandas/pivot-table](https://ds-edu-trainer.vercel.app/reference/pandas/pivot-table) · [pandas/shift-diff](https://ds-edu-trainer.vercel.app/reference/pandas/shift-diff) · [pandas/plot](https://ds-edu-trainer.vercel.app/reference/pandas/plot) · [pandas/cheatsheet](https://ds-edu-trainer.vercel.app/reference/pandas/cheatsheet)"
   ],
   "id": "more"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python"
  },
  "colab": {
   "provenance": []
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}