01_Programming_Historian_ads.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Preparing a computer vision training set for classifying advertisements \n",
    "\n",
    "This notebook covers the steps involved in preparing a labeled dataset of images derived from the Newspaper Navigator dataset. Specifically this training set is primarily intended for use in a Programming Historian lesson on computer vision. \n",
    "\n",
    "## Aims \n",
    "This notebook cover the steps to creating a dataset based on the news-navigator dataset. This includes:\n",
    "- creating a subsample of the news navigator training set\n",
    "- setting up the annotation process\n",
    "- processing the annotations \n",
    "- uploading the dataset to Zenodo\n",
    "\n",
    "It doesn't cover training a model based on this data. This process is covered in a the programming historian lesson for which this dataset was created. \n",
    "\n",
    "## Requirements \n",
    "The easiest way to make sure you have the Python requirements available for this notebook is to install the package via PyPi into a new environment. Instructions can be found on the [README page](https://github.com/davanstrien/nnAnnotator/) for this repository."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importing required libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nnAnnotator import sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import json \n",
    "import requests\n",
    "from pathlib import Path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Set plot style"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.style.use('seaborn')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Working with newspaper navigator dataset\n",
    "The newspaper navigator dataset is available via a few different formats. nnAnnotator uses the JSON files, let's see how to access these. The JSON file is availabl via a URL that looks like this"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'https://news-navigator.labs.loc.gov/prepackaged/1930_ads.json'"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'https://news-navigator.labs.loc.gov/prepackaged/1930_ads.json'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All of these URLS follow the same patern:\n",
    "https://news-navigator.labs.loc.gov/prepackaged/{YEAR}_{KIND}.json As a result it is very simple to grab the correct url for a year and kind of image in the dataset. This is what ```get_json_url``` does: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'https://news-navigator.labs.loc.gov/prepackaged/1940_ads.json'"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample.get_json_url(1940,'ads')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now take a look at an example for the maps data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "url = sample.get_json_url(1900,'maps')\n",
    "with requests.get(url) as r:\n",
    "    data = json.loads(r.content)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Take a look at the first record:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'filepath': 'dlc_alf_ver01/data/sn83045462/00280655351/1900092901/0515/006_2_91.jpg',\n",
       " 'pub_date': '1900-09-29',\n",
       " 'page_seq_num': 515,\n",
       " 'edition_seq_num': 1,\n",
       " 'batch': 'dlc_alf_ver01',\n",
       " 'lccn': 'sn83045462',\n",
       " 'box': [0.7022514105902777,\n",
       "  0.022746477387920072,\n",
       "  0.970748046875,\n",
       "  0.23085515669300366],\n",
       " 'score': 0.917590856552124,\n",
       " 'ocr': ['The',\n",
       "  'Positions',\n",
       "  'of',\n",
       "  'the',\n",
       "  'Principal',\n",
       "  'Stars',\n",
       "  'wbtrh',\n",
       "  'ore',\n",
       "  'above',\n",
       "  'the',\n",
       "  'Horizon',\n",
       "  'October',\n",
       "  '1?15?31,',\n",
       "  'at',\n",
       "  '9',\n",
       "  '8?7',\n",
       "  'p.m.'],\n",
       " 'place_of_publication': 'Washington, D.C.',\n",
       " 'geographic_coverage': ['District of Columbia--Washington'],\n",
       " 'name': 'Evening star. [volume]',\n",
       " 'publisher': 'W.D. Wallach & Hope',\n",
       " 'url': 'https://news-navigator.labs.loc.gov/data/dlc_alf_ver01/data/sn83045462/00280655351/1900092901/0515/006_2_91.jpg',\n",
       " 'page_url': 'https://chroniclingamerica.loc.gov/data/batches/dlc_alf_ver01/data/sn83045462/00280655351/1900092901/0515.jp2'}"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can see there are a bunch of things we might be interested in using either for processing the data or for analysis. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['filepath', 'pub_date', 'page_seq_num', 'edition_seq_num', 'batch', 'lccn', 'box', 'score', 'ocr', 'place_of_publication', 'geographic_coverage', 'name', 'publisher', 'url', 'page_url'])"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[0].keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another notebook will cover the analysis part in more detail, so let's get back to the problem of constructing a dataset. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sampling\n",
    "Since the aim is to create a dataset for training a machine to predict the correct labels for the images we are likely to want a sample of the full dataset rather than the whole thing. Before we decide what sort of sample size might make sense. \n",
    "\n",
    "## View the amount of images for a year\n",
    "Since each record is one image we can easily get the number of images by counting the lenght of a json file for a year and kind:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "https://news-navigator.labs.loc.gov/prepackaged/1900_maps.json has: 1,707 image records\n"
     ]
    }
   ],
   "source": [
    "url = sample.get_json_url(1900,'maps')\n",
    "r = requests.get(url)\n",
    "data = json.loads(r.content)\n",
    "print(f'{url} has: {len(data):,} image records')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Out of memory data \n",
    "This works fine for the maps data but the size of some of the datasets, including those for ads, is too big to load into memory on some machine. Since we want to be able to do our create our sampling on standard device we can use  ```get_year_sizes``` to get the number of records for a type of image and a year. This function avoids loading the full data into memory so will work even if you only have a few GBs of RAM. The [Documentation]() !todo add link has more information on how this works. Let's grab the number of entries for maps:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e351c9c3f66c48ed972805911036dd07",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=11.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "maps = sample.get_year_sizes('maps',step=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can also change the start and end year which default to 1850-1950 or alter the step size. This function returns a dataframe, of counts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>maps_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1850</th>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1860</th>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1870</th>\n",
       "      <td>276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1880</th>\n",
       "      <td>1043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1890</th>\n",
       "      <td>691</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1900</th>\n",
       "      <td>1707</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1910</th>\n",
       "      <td>4979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1920</th>\n",
       "      <td>4744</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1930</th>\n",
       "      <td>3348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1940</th>\n",
       "      <td>2907</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1950</th>\n",
       "      <td>1952</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      maps_count\n",
       "1850           5\n",
       "1860          14\n",
       "1870         276\n",
       "1880        1043\n",
       "1890         691\n",
       "1900        1707\n",
       "1910        4979\n",
       "1920        4744\n",
       "1930        3348\n",
       "1940        2907\n",
       "1950        1952"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "maps"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can see that the number of maps are quite small. Let's now turn to building a dataset for advert type. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Building a training set for classifying advert types\n",
    "\n",
    "We want to build a dataset which labels if an advert image is 'text only' or contains an illustration. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Text only \n",
    "\n",
    "<figure>\n",
    "    <img src=\"figs/text_ad.jpg\" width=\"250\">\n",
    "    <figcaption>Contains only text, in this case we don't count a fancy border as an illustration, or fancy typography which might make a shape but still consits of text.</figcaption>\n",
    "</figure> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Illustrated \n",
    "<figure>\n",
    "    <img src=\"figs/text_ad.jpg\" width=\"250\">\n",
    "    <figcaption>Also contains illustrations only text, again we don't count fancy text as an illustration.</figcaption>\n",
    "</figure> "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading previous counts \n",
    "\n",
    "Because the `get_year_sizes` function requires streaming all the JSON files which contain the newspaper navigator dataset it is fairly slow (depending on how fast your connection is). The `metadata/` folder in this repository already has the year counts. Since the Newspaper Navigator data probably won't change all the time you are probably better to just load this version. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>ads_count</th>\n",
       "      <th>photos_count</th>\n",
       "      <th>maps_count</th>\n",
       "      <th>illustrations_count</th>\n",
       "      <th>comics_count</th>\n",
       "      <th>cartoons_count</th>\n",
       "      <th>headlines_count</th>\n",
       "      <th>total</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1850</th>\n",
       "      <td>8841</td>\n",
       "      <td>22</td>\n",
       "      <td>5</td>\n",
       "      <td>671</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>11243</td>\n",
       "      <td>20791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1851</th>\n",
       "      <td>10065</td>\n",
       "      <td>20</td>\n",
       "      <td>6</td>\n",
       "      <td>457</td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>12262</td>\n",
       "      <td>22817</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1852</th>\n",
       "      <td>8764</td>\n",
       "      <td>22</td>\n",
       "      <td>10</td>\n",
       "      <td>671</td>\n",
       "      <td>10</td>\n",
       "      <td>8</td>\n",
       "      <td>13524</td>\n",
       "      <td>23009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1853</th>\n",
       "      <td>11517</td>\n",
       "      <td>45</td>\n",
       "      <td>5</td>\n",
       "      <td>1106</td>\n",
       "      <td>88</td>\n",
       "      <td>1</td>\n",
       "      <td>13224</td>\n",
       "      <td>25986</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1854</th>\n",
       "      <td>15050</td>\n",
       "      <td>221</td>\n",
       "      <td>15</td>\n",
       "      <td>732</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>15282</td>\n",
       "      <td>31314</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1946</th>\n",
       "      <td>185139</td>\n",
       "      <td>5945</td>\n",
       "      <td>1857</td>\n",
       "      <td>1053</td>\n",
       "      <td>3280</td>\n",
       "      <td>861</td>\n",
       "      <td>68275</td>\n",
       "      <td>266410</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1947</th>\n",
       "      <td>181223</td>\n",
       "      <td>4188</td>\n",
       "      <td>1750</td>\n",
       "      <td>1115</td>\n",
       "      <td>3630</td>\n",
       "      <td>797</td>\n",
       "      <td>57018</td>\n",
       "      <td>249721</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1948</th>\n",
       "      <td>152987</td>\n",
       "      <td>4282</td>\n",
       "      <td>1359</td>\n",
       "      <td>1154</td>\n",
       "      <td>3031</td>\n",
       "      <td>624</td>\n",
       "      <td>43432</td>\n",
       "      <td>206869</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1949</th>\n",
       "      <td>154510</td>\n",
       "      <td>6015</td>\n",
       "      <td>1888</td>\n",
       "      <td>1280</td>\n",
       "      <td>3356</td>\n",
       "      <td>634</td>\n",
       "      <td>42904</td>\n",
       "      <td>210587</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1950</th>\n",
       "      <td>154961</td>\n",
       "      <td>5630</td>\n",
       "      <td>1952</td>\n",
       "      <td>1223</td>\n",
       "      <td>3893</td>\n",
       "      <td>704</td>\n",
       "      <td>37854</td>\n",
       "      <td>206217</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>101 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      ads_count  photos_count  maps_count  illustrations_count  comics_count  \\\n",
       "1850       8841            22           5                  671             9   \n",
       "1851      10065            20           6                  457             7   \n",
       "1852       8764            22          10                  671            10   \n",
       "1853      11517            45           5                 1106            88   \n",
       "1854      15050           221          15                  732            11   \n",
       "...         ...           ...         ...                  ...           ...   \n",
       "1946     185139          5945        1857                 1053          3280   \n",
       "1947     181223          4188        1750                 1115          3630   \n",
       "1948     152987          4282        1359                 1154          3031   \n",
       "1949     154510          6015        1888                 1280          3356   \n",
       "1950     154961          5630        1952                 1223          3893   \n",
       "\n",
       "      cartoons_count  headlines_count   total  \n",
       "1850               0            11243   20791  \n",
       "1851               0            12262   22817  \n",
       "1852               8            13524   23009  \n",
       "1853               1            13224   25986  \n",
       "1854               3            15282   31314  \n",
       "...              ...              ...     ...  \n",
       "1946             861            68275  266410  \n",
       "1947             797            57018  249721  \n",
       "1948             624            43432  206869  \n",
       "1949             634            42904  210587  \n",
       "1950             704            37854  206217  \n",
       "\n",
       "[101 rows x 8 columns]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('metadata/all_year_counts.csv', index_col=0)\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plotting \n",
    "\n",
    "We can use the usual pandas plotting to get an overview of this data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ads as a portion of the total"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can already see that ads make up a large proportion of the total data. Looking at ads and total only gives us a better sense of how these two relate:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[['ads_count','total']].plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Stopping at 1920\n",
    "We can see a steep decline of all types of image in the 1920s."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df[['ads_count','total']][df.index<=1920].plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What does this tell us? \n",
    "\n",
    "This chart shows us a few different things \n",
    "- the number of images predicted as adverts (and the total of all images) predicted by the Newspaper Navigator \n",
    "\n",
    "There are also quite a few things which this doesn't show\n",
    "- the number of adverts grew over time, up until the 1920s. This growth may be explained by more newspapers being published over time, better classifications of adverts by the newspaper navigator algorithm in later years,or a larger body of digitsed material overall in the LOC collection. \n",
    "\n",
    "### Why the drop?\n",
    "The drop in numbers is mostly explained by the copyright status of newspapers published after this period but could also be explained by the Newspaper Navigator classifier performing worse in later years, particularly since the original training data for Newspaper Navigator was trained on WWI data. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Creating a sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Sampling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can use ```sample``` to create samples in a few different ways. Since we're concerned with preparing a sample for annotation we want to limit how much we need to annotate. Based on the visulisations of the growth in the dataset between 1880 and 1900 we want to sample this period. \n",
    "\n",
    "We can use the ```create_sample``` method to create a sample between 1880 and 1900 taking a sample every 5 years. Since our aim is to limit the number of annotation instead of specifcying how much we want to sample from each year we use ```year_sample=False``` to instead for a total sample size ~500. This number isn't precise because the sample size per year is rounded up. \n",
    "\n",
    "This method will return a Pandas Dataframe. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sampler = sample.nnSampler()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2335f2bd65d943ceacc83691faf7720a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=9.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "df = sampler.create_sample(sample_size=500, kind='ads', start_year=1860, end_year=1900, step=5, year_sample=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>filepath</th>\n",
       "      <th>pub_date</th>\n",
       "      <th>page_seq_num</th>\n",
       "      <th>edition_seq_num</th>\n",
       "      <th>batch</th>\n",
       "      <th>lccn</th>\n",
       "      <th>box</th>\n",
       "      <th>score</th>\n",
       "      <th>ocr</th>\n",
       "      <th>place_of_publication</th>\n",
       "      <th>geographic_coverage</th>\n",
       "      <th>name</th>\n",
       "      <th>publisher</th>\n",
       "      <th>url</th>\n",
       "      <th>page_url</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>iahi_gastly_ver01/data/sn82015737/00279529091/...</td>\n",
       "      <td>1860-03-09</td>\n",
       "      <td>447</td>\n",
       "      <td>1</td>\n",
       "      <td>iahi_gastly_ver01</td>\n",
       "      <td>sn82015737</td>\n",
       "      <td>[0.30762831315880534, 0.049735121606933906, 0....</td>\n",
       "      <td>0.950152</td>\n",
       "      <td>[JTO, TMCE, An, t%E, 3eott, County, Pair, is, ...</td>\n",
       "      <td>Davenport, Iowa</td>\n",
       "      <td>[Iowa--Scott--Davenport]</td>\n",
       "      <td>Daily Democrat and news. [volume]</td>\n",
       "      <td>Maguire, Richardson &amp; Co.</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/iahi_...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ohi_cobweb_ver04/data/sn85026050/00280775848/1...</td>\n",
       "      <td>1860-08-17</td>\n",
       "      <td>359</td>\n",
       "      <td>1</td>\n",
       "      <td>ohi_cobweb_ver04</td>\n",
       "      <td>sn85026050</td>\n",
       "      <td>[0.5799164973813336, 0.7307444280707744, 0.700...</td>\n",
       "      <td>0.985859</td>\n",
       "      <td>[9, BI., I, .QJtf, A, never, falllu, Antidote,...</td>\n",
       "      <td>Fremont, Sandusky County [Ohio]</td>\n",
       "      <td>[Ohio--Sandusky--Fremont]</td>\n",
       "      <td>Fremont journal. [volume]</td>\n",
       "      <td>I.W. Booth</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/ohi_c...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>curiv_jojoba_ver01/data/sn84038814/00279557207...</td>\n",
       "      <td>1860-11-22</td>\n",
       "      <td>498</td>\n",
       "      <td>1</td>\n",
       "      <td>curiv_jojoba_ver01</td>\n",
       "      <td>sn84038814</td>\n",
       "      <td>[0.5793982023445715, 0.43399254432091344, 0.70...</td>\n",
       "      <td>0.986739</td>\n",
       "      <td>[]</td>\n",
       "      <td>Marysville, Calif.</td>\n",
       "      <td>[California--Yuba--Marysville]</td>\n",
       "      <td>Daily national Democrat. [volume]</td>\n",
       "      <td>A.S. Randall &amp; Co.</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/curiv...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>hihouml_lilac_ver02/data/sn82015408/0021247138...</td>\n",
       "      <td>1860-03-17</td>\n",
       "      <td>833</td>\n",
       "      <td>1</td>\n",
       "      <td>hihouml_lilac_ver02</td>\n",
       "      <td>sn82015408</td>\n",
       "      <td>[0.5183942753296475, 0.5235361735026042, 0.674...</td>\n",
       "      <td>0.905783</td>\n",
       "      <td>[C., H., L.EWEUS., NOTICE., Competition, the, ...</td>\n",
       "      <td>Honolulu [Oahu], Hawaii</td>\n",
       "      <td>[Hawaii--Honolulu--Honolulu]</td>\n",
       "      <td>Polynesian. [volume]</td>\n",
       "      <td>[publisher not identified]</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/hihou...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>curiv_jojoba_ver01/data/sn84038814/00279557190...</td>\n",
       "      <td>1860-02-21</td>\n",
       "      <td>816</td>\n",
       "      <td>1</td>\n",
       "      <td>curiv_jojoba_ver01</td>\n",
       "      <td>sn84038814</td>\n",
       "      <td>[0.598211174337273, 0.0393043301089006, 0.7273...</td>\n",
       "      <td>0.918803</td>\n",
       "      <td>[]</td>\n",
       "      <td>Marysville, Calif.</td>\n",
       "      <td>[California--Yuba--Marysville]</td>\n",
       "      <td>Daily national Democrat. [volume]</td>\n",
       "      <td>A.S. Randall &amp; Co.</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/curiv...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>wvu_austria_ver01/data/sn84026845/00202190650/...</td>\n",
       "      <td>1860-03-22</td>\n",
       "      <td>277</td>\n",
       "      <td>1</td>\n",
       "      <td>wvu_austria_ver01</td>\n",
       "      <td>sn84026845</td>\n",
       "      <td>[0.2810836642761977, 0.021020836717500463, 0.5...</td>\n",
       "      <td>0.960305</td>\n",
       "      <td>[B., A., Fahnestock, &amp;, Co's., PURE, WHITE, LE...</td>\n",
       "      <td>Wheeling, Va. [W. Va.]</td>\n",
       "      <td>[West Virginia--Ohio--Wheeling]</td>\n",
       "      <td>Daily intelligencer. [volume]</td>\n",
       "      <td>Campbell &amp; M'Dermot</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/wvu_a...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>tu_furry_ver01/data/sn83045628/00200293587/186...</td>\n",
       "      <td>1860-04-21</td>\n",
       "      <td>69</td>\n",
       "      <td>1</td>\n",
       "      <td>tu_furry_ver01</td>\n",
       "      <td>sn83045628</td>\n",
       "      <td>[0.2715182629933798, 0.5381755536703364, 0.388...</td>\n",
       "      <td>0.943606</td>\n",
       "      <td>[IL'J'S, ami, (, 'ATS,, BONNETS, till, OOODS,,...</td>\n",
       "      <td>Knoxville, Tenn.</td>\n",
       "      <td>[Tennessee--Knox--Knoxville]</td>\n",
       "      <td>Brownlow's Knoxville Whig. [volume]</td>\n",
       "      <td>E.B.P. Kinsloe</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/tu_fu...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>iahi_bellsprout_ver01/data/sn84027352/00279528...</td>\n",
       "      <td>1860-02-02</td>\n",
       "      <td>444</td>\n",
       "      <td>1</td>\n",
       "      <td>iahi_bellsprout_ver01</td>\n",
       "      <td>sn84027352</td>\n",
       "      <td>[0.5682708976804748, 0.5028207913385059, 0.703...</td>\n",
       "      <td>0.938462</td>\n",
       "      <td>['Vimilli, 1, i, (FALL, &amp;, WINTER, Mammoth, Cl...</td>\n",
       "      <td>Ottumwa, Iowa</td>\n",
       "      <td>[Iowa--Wapello--Ottumwa]</td>\n",
       "      <td>The weekly Ottumwa courier. [volume]</td>\n",
       "      <td>J.W. &amp; G.P. Norris</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/iahi_...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>curiv_jojoba_ver01/data/sn84038814/00279557207...</td>\n",
       "      <td>1860-07-26</td>\n",
       "      <td>87</td>\n",
       "      <td>1</td>\n",
       "      <td>curiv_jojoba_ver01</td>\n",
       "      <td>sn84038814</td>\n",
       "      <td>[0.6939450665470904, 0.29700505687745826, 0.81...</td>\n",
       "      <td>0.959135</td>\n",
       "      <td>[]</td>\n",
       "      <td>Marysville, Calif.</td>\n",
       "      <td>[California--Yuba--Marysville]</td>\n",
       "      <td>Daily national Democrat. [volume]</td>\n",
       "      <td>A.S. Randall &amp; Co.</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/curiv...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>ohi_arnarson_ver01/data/sn83035173/00296027297...</td>\n",
       "      <td>1860-12-12</td>\n",
       "      <td>412</td>\n",
       "      <td>1</td>\n",
       "      <td>ohi_arnarson_ver01</td>\n",
       "      <td>sn83035173</td>\n",
       "      <td>[0.053070784451668726, 0.656615614353266, 0.19...</td>\n",
       "      <td>0.946558</td>\n",
       "      <td>[JVC, nil, whrm, if, mjiv, V.cwxoptw, !, m, JL...</td>\n",
       "      <td>Ashland, Ashland County, Ohio</td>\n",
       "      <td>[Ohio--Ashland--Ashland]</td>\n",
       "      <td>The Ashland union. [volume]</td>\n",
       "      <td>Jno. Sheridan</td>\n",
       "      <td>https://news-navigator.labs.loc.gov/data/ohi_a...</td>\n",
       "      <td>https://chroniclingamerica.loc.gov/data/batche...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            filepath    pub_date  \\\n",
       "0  iahi_gastly_ver01/data/sn82015737/00279529091/...  1860-03-09   \n",
       "1  ohi_cobweb_ver04/data/sn85026050/00280775848/1...  1860-08-17   \n",
       "2  curiv_jojoba_ver01/data/sn84038814/00279557207...  1860-11-22   \n",
       "3  hihouml_lilac_ver02/data/sn82015408/0021247138...  1860-03-17   \n",
       "4  curiv_jojoba_ver01/data/sn84038814/00279557190...  1860-02-21   \n",
       "5  wvu_austria_ver01/data/sn84026845/00202190650/...  1860-03-22   \n",
       "6  tu_furry_ver01/data/sn83045628/00200293587/186...  1860-04-21   \n",
       "7  iahi_bellsprout_ver01/data/sn84027352/00279528...  1860-02-02   \n",
       "8  curiv_jojoba_ver01/data/sn84038814/00279557207...  1860-07-26   \n",
       "9  ohi_arnarson_ver01/data/sn83035173/00296027297...  1860-12-12   \n",
       "\n",
       "   page_seq_num edition_seq_num                  batch        lccn  \\\n",
       "0           447               1      iahi_gastly_ver01  sn82015737   \n",
       "1           359               1       ohi_cobweb_ver04  sn85026050   \n",
       "2           498               1     curiv_jojoba_ver01  sn84038814   \n",
       "3           833               1    hihouml_lilac_ver02  sn82015408   \n",
       "4           816               1     curiv_jojoba_ver01  sn84038814   \n",
       "5           277               1      wvu_austria_ver01  sn84026845   \n",
       "6            69               1         tu_furry_ver01  sn83045628   \n",
       "7           444               1  iahi_bellsprout_ver01  sn84027352   \n",
       "8            87               1     curiv_jojoba_ver01  sn84038814   \n",
       "9           412               1     ohi_arnarson_ver01  sn83035173   \n",
       "\n",
       "                                                 box     score  \\\n",
       "0  [0.30762831315880534, 0.049735121606933906, 0....  0.950152   \n",
       "1  [0.5799164973813336, 0.7307444280707744, 0.700...  0.985859   \n",
       "2  [0.5793982023445715, 0.43399254432091344, 0.70...  0.986739   \n",
       "3  [0.5183942753296475, 0.5235361735026042, 0.674...  0.905783   \n",
       "4  [0.598211174337273, 0.0393043301089006, 0.7273...  0.918803   \n",
       "5  [0.2810836642761977, 0.021020836717500463, 0.5...  0.960305   \n",
       "6  [0.2715182629933798, 0.5381755536703364, 0.388...  0.943606   \n",
       "7  [0.5682708976804748, 0.5028207913385059, 0.703...  0.938462   \n",
       "8  [0.6939450665470904, 0.29700505687745826, 0.81...  0.959135   \n",
       "9  [0.053070784451668726, 0.656615614353266, 0.19...  0.946558   \n",
       "\n",
       "                                                 ocr  \\\n",
       "0  [JTO, TMCE, An, t%E, 3eott, County, Pair, is, ...   \n",
       "1  [9, BI., I, .QJtf, A, never, falllu, Antidote,...   \n",
       "2                                                 []   \n",
       "3  [C., H., L.EWEUS., NOTICE., Competition, the, ...   \n",
       "4                                                 []   \n",
       "5  [B., A., Fahnestock, &, Co's., PURE, WHITE, LE...   \n",
       "6  [IL'J'S, ami, (, 'ATS,, BONNETS, till, OOODS,,...   \n",
       "7  ['Vimilli, 1, i, (FALL, &, WINTER, Mammoth, Cl...   \n",
       "8                                                 []   \n",
       "9  [JVC, nil, whrm, if, mjiv, V.cwxoptw, !, m, JL...   \n",
       "\n",
       "              place_of_publication              geographic_coverage  \\\n",
       "0                  Davenport, Iowa         [Iowa--Scott--Davenport]   \n",
       "1  Fremont, Sandusky County [Ohio]        [Ohio--Sandusky--Fremont]   \n",
       "2               Marysville, Calif.   [California--Yuba--Marysville]   \n",
       "3          Honolulu [Oahu], Hawaii     [Hawaii--Honolulu--Honolulu]   \n",
       "4               Marysville, Calif.   [California--Yuba--Marysville]   \n",
       "5           Wheeling, Va. [W. Va.]  [West Virginia--Ohio--Wheeling]   \n",
       "6                 Knoxville, Tenn.     [Tennessee--Knox--Knoxville]   \n",
       "7                    Ottumwa, Iowa         [Iowa--Wapello--Ottumwa]   \n",
       "8               Marysville, Calif.   [California--Yuba--Marysville]   \n",
       "9    Ashland, Ashland County, Ohio         [Ohio--Ashland--Ashland]   \n",
       "\n",
       "                                   name                   publisher  \\\n",
       "0     Daily Democrat and news. [volume]   Maguire, Richardson & Co.   \n",
       "1             Fremont journal. [volume]                  I.W. Booth   \n",
       "2     Daily national Democrat. [volume]          A.S. Randall & Co.   \n",
       "3                  Polynesian. [volume]  [publisher not identified]   \n",
       "4     Daily national Democrat. [volume]          A.S. Randall & Co.   \n",
       "5         Daily intelligencer. [volume]         Campbell & M'Dermot   \n",
       "6   Brownlow's Knoxville Whig. [volume]              E.B.P. Kinsloe   \n",
       "7  The weekly Ottumwa courier. [volume]          J.W. & G.P. Norris   \n",
       "8     Daily national Democrat. [volume]          A.S. Randall & Co.   \n",
       "9           The Ashland union. [volume]               Jno. Sheridan   \n",
       "\n",
       "                                                 url  \\\n",
       "0  https://news-navigator.labs.loc.gov/data/iahi_...   \n",
       "1  https://news-navigator.labs.loc.gov/data/ohi_c...   \n",
       "2  https://news-navigator.labs.loc.gov/data/curiv...   \n",
       "3  https://news-navigator.labs.loc.gov/data/hihou...   \n",
       "4  https://news-navigator.labs.loc.gov/data/curiv...   \n",
       "5  https://news-navigator.labs.loc.gov/data/wvu_a...   \n",
       "6  https://news-navigator.labs.loc.gov/data/tu_fu...   \n",
       "7  https://news-navigator.labs.loc.gov/data/iahi_...   \n",
       "8  https://news-navigator.labs.loc.gov/data/curiv...   \n",
       "9  https://news-navigator.labs.loc.gov/data/ohi_a...   \n",
       "\n",
       "                                            page_url  \n",
       "0  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "1  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "2  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "3  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "4  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "5  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "6  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "7  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "8  https://chroniclingamerica.loc.gov/data/batche...  \n",
       "9  https://chroniclingamerica.loc.gov/data/batche...  "
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save the output.\n",
    "\n",
    "Let's make a directory to save all of the outsputs of this dataset. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Path('ph/ads').mkdir(parents=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sampler.sample.to_csv('ph/ads/sample.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's check the number per year:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1885    61\n",
       "1880    61\n",
       "1870    61\n",
       "1890    61\n",
       "1895    61\n",
       "1860    61\n",
       "1875    61\n",
       "1900    61\n",
       "1865    61\n",
       "Name: pub_date, dtype: int64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['pub_date'].map(lambda x: x.split('-')[0]).value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can see that the function used to create a sample created even samples for each year. The intution here is that since we want a classifier that performs equally well on each year we want to make sure that the data is the same for each year. This could be the wrong approach, it might be better to recreate a dataset which is proprtionally sampled for each year. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Other info in the dataset\n",
    "Let's take a look at a few other bits of the data "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[publisher not identified]     17\n",
       "Bingham & Doughty               9\n",
       "Douglass & Conner               8\n",
       "N.A. Foster & Co.               8\n",
       "New York Tribune                8\n",
       "S.T. Seawell & W.N. Stanton     7\n",
       "J.B. Howell                     7\n",
       "Maguire, Richardson & Co.       7\n",
       "Post-Intelligencer Pub. Co.     6\n",
       "Edward Rosewater                6\n",
       "John H. Oberly & Co.            6\n",
       "Edgar Snowden                   6\n",
       "S. Medary                       6\n",
       "Smith, Camp & Co.               5\n",
       "W. & G. Dunbar, Jr.             5\n",
       "Name: publisher, dtype: int64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['publisher'].value_counts().head(15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Average number of photos per month'}>"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['pub_date'].map(lambda x: x.split('-')[1]).value_counts(normalize=True).sort_index().plot(kind='bar', title='Average number of photos per month')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['month'] = df['pub_date'].map(lambda x: x.split('-')[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['year'] = df['pub_date'].map(lambda x:x.split('-')[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "grouped = df.groupby(['year'])['month'].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "year  month\n",
       "1860  05       0.163934\n",
       "      04       0.131148\n",
       "      07       0.114754\n",
       "      02       0.098361\n",
       "      08       0.098361\n",
       "                 ...   \n",
       "1900  10       0.081967\n",
       "      01       0.065574\n",
       "      04       0.049180\n",
       "      07       0.049180\n",
       "      03       0.032787\n",
       "Name: month, Length: 107, dtype: float64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Download sample \n",
    "\n",
    "To download the actual images we use `download_sample`, passing in a folder to save the output images. In this case since we already have a sample dataframe this is passed in."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d11bc53181714978869302a9cd2a4ea0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=549.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "sampler.download_sample('ph/ads/images',pct=80)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Setup Label Studio\n",
    "\n",
    "Label studio can be used to create annotation tasks in a few different ways. In this case it is done via the CLI. We can navigate to the images folder and run `label-studio` init to setup the annotation task. The other flags are hopefully fairly clear. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      https://chroniclingamerica.loc.gov/iiif/2/iahi...\n",
       "1      https://chroniclingamerica.loc.gov/iiif/2/ohi_...\n",
       "2      https://chroniclingamerica.loc.gov/iiif/2/curi...\n",
       "3      https://chroniclingamerica.loc.gov/iiif/2/hiho...\n",
       "4      https://chroniclingamerica.loc.gov/iiif/2/curi...\n",
       "                             ...                        \n",
       "544    https://chroniclingamerica.loc.gov/iiif/2/wa_i...\n",
       "545    https://chroniclingamerica.loc.gov/iiif/2/nbu_...\n",
       "546    https://chroniclingamerica.loc.gov/iiif/2/kyu_...\n",
       "547    https://chroniclingamerica.loc.gov/iiif/2/ct_h...\n",
       "548    https://chroniclingamerica.loc.gov/iiif/2/nbu_...\n",
       "Name: iiif_url, Length: 549, dtype: object"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "df = pd.read_csv('ph/ads/images/2020_08_10_19_31_549_sample.csv',index_col=0)\n",
    "df['iiif_url']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```bash\n",
    "cd /ph/ads/\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```bash\n",
    "label-studio init ad_annotations --template=image_classification\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```bash\n",
    "label-studio start ./ad_annotations\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the config for the annotation task\n",
    "\n",
    "Once you are in the label studio interface you will be prompted to define an annotation task, this can be done using an XML like template. For this task the following was used. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```html\n",
    "<View>\n",
    "  <Image name=\"image\" value=\"$image\"></Image>\n",
    "  <Choices name=\"choice\" toName=\"image\" showInLine=\"true\">\n",
    "    <Choice value=\"text-only\" background=\"blue\"></Choice>\n",
    "    <Choice value=\"illustrations\" background=\"green\"></Choice>\n",
    "  </Choices>\n",
    "</View>\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Annotate\n",
    "Once you have loaded your images and defined the labels you can start annotating. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Process annotations\n",
    "Once you have finished annotating, or are too bored to carry on you can use functionality from `annotate` to help process the annotations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nnAnnotator import annotate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can check what kind of information is stored with our annotations. These consist of various configuration files, and the 'completions' which are single json files which store completed annotations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m\u001b[36mcompletions\u001b[m\u001b[m config.json \u001b[31mconfig.xml\u001b[m\u001b[m  tasks.json\n"
     ]
    }
   ],
   "source": [
    "!ls ph/ads/ad_annotations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "completions\n",
      "config.json\n",
      "config.xml\n",
      "tasks.json\n"
     ]
    }
   ],
   "source": [
    "!ls ph/ads/ad_annotations | head -n 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "annotations = annotate.nnAnnotations.from_completions('ph/ads/ad_annotations', 'classification')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`process_completions` can be used to put these completed annotations into a dataframe. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>created_at</th>\n",
       "      <th>id</th>\n",
       "      <th>lead_time</th>\n",
       "      <th>result</th>\n",
       "      <th>data</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1602236711</td>\n",
       "      <td>379001</td>\n",
       "      <td>1.248</td>\n",
       "      <td>text-only</td>\n",
       "      <td>pst_fenske_ver02_data_sn84026497_00280776129_1...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1602237071</td>\n",
       "      <td>396001</td>\n",
       "      <td>0.870</td>\n",
       "      <td>text-only</td>\n",
       "      <td>scu_carlacox_ver01_data_sn84026965_00294551268...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1602237268</td>\n",
       "      <td>115001</td>\n",
       "      <td>1.949</td>\n",
       "      <td>text-only</td>\n",
       "      <td>in_england_ver02_data_sn82015464_00202191435_1...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1602236719</td>\n",
       "      <td>545001</td>\n",
       "      <td>0.932</td>\n",
       "      <td>text-only</td>\n",
       "      <td>wvu_isner_ver01_data_sn86092518_0041418683A_18...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1602237545</td>\n",
       "      <td>400001</td>\n",
       "      <td>0.775</td>\n",
       "      <td>text-only</td>\n",
       "      <td>scu_drinksmall_ver01_data_sn93067705_002372884...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>547</th>\n",
       "      <td>1602237030</td>\n",
       "      <td>244001</td>\n",
       "      <td>1.069</td>\n",
       "      <td>illustrations</td>\n",
       "      <td>mnhi_edina_ver01_data_sn90059522_0017503349A_1...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>548</th>\n",
       "      <td>1602237445</td>\n",
       "      <td>497001</td>\n",
       "      <td>0.785</td>\n",
       "      <td>text-only</td>\n",
       "      <td>vtu_burlington_ver01_data_sn86072143_002807770...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>549</th>\n",
       "      <td>1602236821</td>\n",
       "      <td>35001</td>\n",
       "      <td>0.780</td>\n",
       "      <td>text-only</td>\n",
       "      <td>deu_arden_ver01_data_sn85042354_00383343367_18...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>550</th>\n",
       "      <td>1602237006</td>\n",
       "      <td>182001</td>\n",
       "      <td>1.161</td>\n",
       "      <td>illustrations</td>\n",
       "      <td>kyu_dracula_ver01_data_sn85052141_0020653364A_...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>551</th>\n",
       "      <td>1602237571</td>\n",
       "      <td>478001</td>\n",
       "      <td>2.468</td>\n",
       "      <td>illustrations</td>\n",
       "      <td>uuml_hornacek_ver01_data_sn83045555_206539732_...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>549 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     created_at      id  lead_time         result  \\\n",
       "0    1602236711  379001      1.248      text-only   \n",
       "1    1602237071  396001      0.870      text-only   \n",
       "2    1602237268  115001      1.949      text-only   \n",
       "3    1602236719  545001      0.932      text-only   \n",
       "4    1602237545  400001      0.775      text-only   \n",
       "..          ...     ...        ...            ...   \n",
       "547  1602237030  244001      1.069  illustrations   \n",
       "548  1602237445  497001      0.785      text-only   \n",
       "549  1602236821   35001      0.780      text-only   \n",
       "550  1602237006  182001      1.161  illustrations   \n",
       "551  1602237571  478001      2.468  illustrations   \n",
       "\n",
       "                                                  data  \n",
       "0    pst_fenske_ver02_data_sn84026497_00280776129_1...  \n",
       "1    scu_carlacox_ver01_data_sn84026965_00294551268...  \n",
       "2    in_england_ver02_data_sn82015464_00202191435_1...  \n",
       "3    wvu_isner_ver01_data_sn86092518_0041418683A_18...  \n",
       "4    scu_drinksmall_ver01_data_sn93067705_002372884...  \n",
       "..                                                 ...  \n",
       "547  mnhi_edina_ver01_data_sn90059522_0017503349A_1...  \n",
       "548  vtu_burlington_ver01_data_sn86072143_002807770...  \n",
       "549  deu_arden_ver01_data_sn85042354_00383343367_18...  \n",
       "550  kyu_dracula_ver01_data_sn85052141_0020653364A_...  \n",
       "551  uuml_hornacek_ver01_data_sn83045555_206539732_...  \n",
       "\n",
       "[549 rows x 5 columns]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "annotations.annotation_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next we can export these annotations to a csv file. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "annotations.export_annotations('ph/ads/ads.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we have a stripped down version of the dataframe which just contains the file and the labels. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>file</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>pst_fenske_ver02_data_sn84026497_00280776129_1...</td>\n",
       "      <td>text-only</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>scu_carlacox_ver01_data_sn84026965_00294551268...</td>\n",
       "      <td>text-only</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>in_england_ver02_data_sn82015464_00202191435_1...</td>\n",
       "      <td>text-only</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>wvu_isner_ver01_data_sn86092518_0041418683A_18...</td>\n",
       "      <td>text-only</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                file      label\n",
       "0  pst_fenske_ver02_data_sn84026497_00280776129_1...  text-only\n",
       "1  scu_carlacox_ver01_data_sn84026965_00294551268...  text-only\n",
       "2  in_england_ver02_data_sn82015464_00202191435_1...  text-only\n",
       "3  wvu_isner_ver01_data_sn86092518_0041418683A_18...  text-only"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('ph/ads/ads.csv')\n",
    "df.head(4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Upsample minority classes\n",
    "\n",
    "We can see how our labels are distributed using the pandas `value_counts` method. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "text-only        0.684882\n",
       "illustrations    0.315118\n",
       "Name: label, dtype: float64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['label'].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see our classes are imbalanced here. Normally, it would be better to maintain the same balance of lablels. Because this dataset will be used for a tutorial I also creat a balanced version of the dataset by upsampling the data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_majority = df[df.label=='text-only']\n",
    "df_minority = df[df.label=='illustrations']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(376, 173)"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_majority), len(df_minority)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.utils import resample\n",
    "# Upsample minority class\n",
    "df_minority_upsampled = resample(df_minority, \n",
    "                                 replace=True,     # sample with replacement\n",
    "                                 n_samples=len(df_majority),  # to match majority class\n",
    "                                 random_state=42) # reproducible results\n",
    "# Combine majority class with upsampled minority class\n",
    "df_upsampled = pd.concat([df_majority, df_minority_upsampled])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "text-only        376\n",
       "illustrations    376\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_upsampled['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_upsampled.to_csv('ph/ads/ads_upsampled.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample_df = pd.read_csv('ph/ads/images/2020_08_10_19_31_549_sample.csv', index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "annotations.merge_sample(sample_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "annotations.merged_df.to_csv('ph/ads/annotated_sample.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Uploading to Zenodo \n",
    "Now we have processed the annotations we will upload them to Zenodo. We can do this using the zenodo api. To use this API you will need to create an access token. You can find documentation for the API on the [developer section](https://developers.zenodo.org/) of the Zenodo website. \n",
    "\n",
    "dotenv is used to load the access token from an `.env` file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from dotenv import load_dotenv\n",
    "import os\n",
    "load_dotenv()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ACCESS_TOKEN = os.getenv(\"ACCESS_TOKEN\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use requests to upload the data, and tqdm to get a progress bar for later. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests \n",
    "from tqdm.notebook import tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The next steps are \\ # TODO finalize outline "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "r = requests.get('https://zenodo.org/api/deposit/depositions',params={'access_token': ACCESS_TOKEN})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "200"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r.status_code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "201"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "headers = {\"Content-Type\": \"application/json\"}\n",
    "params={'access_token': ACCESS_TOKEN}\n",
    "r = requests.post('https://zenodo.org/api/deposit/depositions',\n",
    "                   params=params,\n",
    "                   json={},\n",
    "                   headers=headers)\n",
    "r.status_code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'conceptrecid': '4075210',\n",
       " 'created': '2020-10-09T13:38:48.393742+00:00',\n",
       " 'files': [],\n",
       " 'id': 4075211,\n",
       " 'links': {'bucket': 'https://zenodo.org/api/files/d379d9e9-9e7c-4d9b-8cb0-365677c567ed',\n",
       "  'discard': 'https://zenodo.org/api/deposit/depositions/4075211/actions/discard',\n",
       "  'edit': 'https://zenodo.org/api/deposit/depositions/4075211/actions/edit',\n",
       "  'files': 'https://zenodo.org/api/deposit/depositions/4075211/files',\n",
       "  'html': 'https://zenodo.org/deposit/4075211',\n",
       "  'latest_draft': 'https://zenodo.org/api/deposit/depositions/4075211',\n",
       "  'latest_draft_html': 'https://zenodo.org/deposit/4075211',\n",
       "  'publish': 'https://zenodo.org/api/deposit/depositions/4075211/actions/publish',\n",
       "  'self': 'https://zenodo.org/api/deposit/depositions/4075211'},\n",
       " 'metadata': {'prereserve_doi': {'doi': '10.5281/zenodo.4075211',\n",
       "   'recid': 4075211}},\n",
       " 'modified': '2020-10-09T13:38:48.393755+00:00',\n",
       " 'owner': 32477,\n",
       " 'record_id': 4075211,\n",
       " 'state': 'unsubmitted',\n",
       " 'submitted': False,\n",
       " 'title': ''}"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r.json()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "bucket_url = r.json()[\"links\"][\"bucket\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once we have a bucket url we are ready to upload our files, in this case we filter out `DS_Store` and upload everything else. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PosixPath('ph/ads/ad_annotations'), PosixPath('ph/ads/images')]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dirs = [x for x in Path('ph/ads').iterdir() if x.is_dir() and not x.name.startswith('.')]\n",
    "dirs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "!zip -j -r -q ph/ads/images.zip ph/ads/images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[PosixPath('ph/ads/images.zip'),\n",
       " PosixPath('ph/ads/annotated_sample.csv'),\n",
       " PosixPath('ph/ads/sample.csv'),\n",
       " PosixPath('ph/ads/ads.csv'),\n",
       " PosixPath('ph/ads/ads_upsampled.csv')]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files = [f for f in Path('ph/ads').iterdir() if f.is_file() and not f.name.startswith('.')]\n",
    "files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "65991ed752a34d1894c857f5942777d3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=5.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'mimetype': 'text/csv',\n",
       " 'updated': '2020-10-09T13:40:29.224322+00:00',\n",
       " 'links': {'self': 'https://zenodo.org/api/files/d379d9e9-9e7c-4d9b-8cb0-365677c567ed/ph/ads/ads_upsampled.csv',\n",
       "  'version': 'https://zenodo.org/api/files/d379d9e9-9e7c-4d9b-8cb0-365677c567ed/ph/ads/ads_upsampled.csv?versionId=e7313ede-d5ac-4678-86f7-151ecc715073',\n",
       "  'uploads': 'https://zenodo.org/api/files/d379d9e9-9e7c-4d9b-8cb0-365677c567ed/ph/ads/ads_upsampled.csv?uploads'},\n",
       " 'is_head': True,\n",
       " 'created': '2020-10-09T13:40:29.218245+00:00',\n",
       " 'checksum': 'md5:547c8245118bc61f30d318bb9ec9d19d',\n",
       " 'version_id': 'e7313ede-d5ac-4678-86f7-151ecc715073',\n",
       " 'delete_marker': False,\n",
       " 'key': 'ph/ads/ads_upsampled.csv',\n",
       " 'size': 68332}"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for file in tqdm(files):\n",
    "    with open(file,\"rb\") as fp:\n",
    "        r = requests.put(\n",
    "            \"%s/%s\" % (bucket_url, file),\n",
    "            data=fp,\n",
    "            params=params)\n",
    "r.json()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once this has been done we should make sure we create an additional metadata related to the upload within Zenodo. "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}