Add notebook generating the updated rare species catalog, lisences, and manifest files

eol_realign/notebooks/re-gen_rs_catalog.ipynb

@@ -0,0 +1,712 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import pandas as pd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "RS_FILEPATH = \"../../rare_species/data/\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Read in Complete Rare Species Manifest"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "<class 'pandas.core.frame.DataFrame'>\n",
+      "RangeIndex: 22146 entries, 0 to 22145\n",
+      "Data columns (total 11 columns):\n",
+      " #   Column                  Non-Null Count  Dtype \n",
+      "---  ------                  --------------  ----- \n",
+      " 0   filename                22146 non-null  object\n",
+      " 1   md5                     22146 non-null  object\n",
+      " 2   combined_id             22146 non-null  object\n",
+      " 3   eol_content_id          22146 non-null  int64 \n",
+      " 4   eol_page_id             22146 non-null  int64 \n",
+      " 5   medium_source_url       22146 non-null  object\n",
+      " 6   eol_full_size_copy_url  22146 non-null  object\n",
+      " 7   license_name            22146 non-null  object\n",
+      " 8   copyright_owner         22146 non-null  object\n",
+      " 9   title                   22146 non-null  object\n",
+      " 10  license_link            22103 non-null  object\n",
+      "dtypes: int64(2), object(9)\n",
+      "memory usage: 1.9+ MB\n"
+     ]
+    }
+   ],
+   "source": [
+    "rs_manifest = pd.read_csv(\"../data/rarespecies_full_manifest.csv\",\n",
+    "                          low_memory=False,\n",
+    "                          dtype = {\"eol_content_id\": \"int64\", \"eol_page_id\": \"int64\"})\n",
+    "rs_manifest.info(show_counts = True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Read in Rare Species Catalog"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Note that the original [`rarespecies-catalog`](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/rare_species/data/rarespecies-catalog.csv) generation was done in [`rare_species_make_catalog.ipynb`](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/rare_species/notebooks/rare_species_make_catalog.ipynb). It used [`predicted-catalog.csv`](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/data/predicted-catalog.csv) to get the taxonomic information and align the [rare species cargo](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/rare_species/data/rarespecies.csv) appropriately. There were also checks that the unseen subset was indeed excluded from the [catalog](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/data/catalog.csv).\n",
+    "\n",
+    "We will use the [`rarespecies-catalog.csv`](https://huggingface.co/datasets/imageomics/ToL-EDA/blob/5db838a9db980f4cf96027b326b06e0f42df0c65/rare_species/data/rarespecies-catalog.csv) generated in that notebook to align taxonomic information with the full rare species manifest (renamed locally to `rarespecies-catalog-orig.csv`)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "<class 'pandas.core.frame.DataFrame'>\n",
+      "RangeIndex: 12000 entries, 0 to 11999\n",
+      "Data columns (total 11 columns):\n",
+      " #   Column          Non-Null Count  Dtype \n",
+      "---  ------          --------------  ----- \n",
+      " 0   rarespecies_id  12000 non-null  object\n",
+      " 1   eol_content_id  12000 non-null  int64 \n",
+      " 2   eol_page_id     12000 non-null  int64 \n",
+      " 3   kingdom         12000 non-null  object\n",
+      " 4   phylum          12000 non-null  object\n",
+      " 5   class           12000 non-null  object\n",
+      " 6   order           12000 non-null  object\n",
+      " 7   family          12000 non-null  object\n",
+      " 8   genus           12000 non-null  object\n",
+      " 9   species         12000 non-null  object\n",
+      " 10  sciName         12000 non-null  object\n",
+      "dtypes: int64(2), object(9)\n",
+      "memory usage: 1.0+ MB\n"
+     ]
+    }
+   ],
+   "source": [
+    "rs_catalog = pd.read_csv(RS_FILEPATH + \"rarespecies-catalog-orig.csv\",\n",
+    "                         low_memory=False,\n",
+    "                         dtype = {\"eol_content_id\": \"int64\", \"eol_page_id\": \"int64\"})\n",
+    "rs_catalog.info(show_counts = True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Check Number of Images per Page"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "22146"
+      ]
+     },
+     "execution_count": 5,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "rs_manifest[\"md5\"].nunique()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "We should have at least 30 images per species, except for 5. Let's check the number of images for each species (all now included are unique, as demonstrated above)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {
+    "lines_to_next_cell": 2
+   },
+   "outputs": [],
+   "source": [
+    "num_images_by_pg = {}\n",
+    "\n",
+    "for pg_id in list(rs_manifest.eol_page_id.unique()):\n",
+    "    num_images = rs_manifest.loc[rs_manifest[\"eol_page_id\"] == pg_id].shape[0]\n",
+    "    num_images_by_pg[pg_id] = num_images\n",
+    "    rs_manifest.loc[rs_manifest[\"eol_page_id\"] == pg_id, \"num_images\"] = num_images\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "93"
+      ]
+     },
+     "execution_count": 7,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "rs_manifest.num_images.nunique()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "There's a lot of variety in the number of images per page. We'll check again on the (expected 5) pages that have less than 30 images."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "page ID 485420 has less than 30 unique images; it has 24 unique images\n",
+      "page ID 914531 has less than 30 unique images; it has 29 unique images\n",
+      "page ID 791049 has less than 30 unique images; it has 27 unique images\n",
+      "page ID 205714 has less than 30 unique images; it has 29 unique images\n",
+      "page ID 1286909 has less than 30 unique images; it has 27 unique images\n"
+     ]
+    }
+   ],
+   "source": [
+    "for pg_id in num_images_by_pg.keys():\n",
+    "    if num_images_by_pg[pg_id] < 30:\n",
+    "        print(f\"page ID {pg_id} has less than 30 unique images; it has {num_images_by_pg[pg_id]} unique images\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "https://eol.org/pages/485420/media has only 21 true images, since 3 are post-its. These will need to be manually noted and removed."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Reduce to 30 Images per Page"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Now we need to reduce to just 30 images per page. We can realign the labels based on the page IDs from the `rarespecies_catalog`, but I won't be able to properly rematch the `rarespecies_id`, so that will need to be regenerated..."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "22146"
+      ]
+     },
+     "execution_count": 9,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# Is filename unique? --yes, 22146 is number of images in the full rarespecies manifest\n",
+    "rs_manifest.filename.nunique()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "keep\n",
+       "yes    11986\n",
+       "Name: count, dtype: int64"
+      ]
+     },
+     "execution_count": 15,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "img_list_by_page = {}\n",
+    "\n",
+    "for page_id in list(rs_manifest.eol_page_id.unique()):\n",
+    "    temp = rs_manifest.loc[rs_manifest.eol_page_id == page_id]\n",
+    "    if temp.num_images.values[0] > 30:\n",
+    "        imgs_to_keep = list(temp.filename.sample(30, random_state = 614)) #seed for reproducibility\n",
+    "    else:\n",
+    "        imgs_to_keep = list(temp.filename)\n",
+    "\n",
+    "    rs_manifest.loc[rs_manifest[\"filename\"].isin(imgs_to_keep), \"keep\"] = \"yes\"\n",
+    "    img_list_by_page[page_id] = imgs_to_keep\n",
+    "\n",
+    "rs_manifest[\"keep\"].value_counts()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "That's 14 less than 12000 images, which matches with the number of images we'd be reduced by due to the 5 pages that have less than 30 images."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "filename                  11986\n",
+       "md5                       11986\n",
+       "combined_id               11986\n",
+       "eol_content_id            11986\n",
+       "eol_page_id                 400\n",
+       "medium_source_url         11986\n",
+       "eol_full_size_copy_url    11986\n",
+       "license_name                 15\n",
+       "copyright_owner            4164\n",
+       "title                      1290\n",
+       "license_link                 10\n",
+       "num_images                   93\n",
+       "keep                          1\n",
+       "dtype: int64"
+      ]
+     },
+     "execution_count": 16,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "reduced_rs_manifest = rs_manifest.loc[rs_manifest[\"keep\"] == \"yes\"]\n",
+    "\n",
+    "reduced_rs_manifest.nunique()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "num_images\n",
+       "30.0    11850\n",
+       "29.0       58\n",
+       "27.0       54\n",
+       "24.0       24\n",
+       "Name: count, dtype: int64"
+      ]
+     },
+     "execution_count": 17,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "for pg_id in list(reduced_rs_manifest.eol_page_id.unique()):\n",
+    "    num_images = reduced_rs_manifest.loc[reduced_rs_manifest[\"eol_page_id\"] == pg_id].shape[0]\n",
+    "    reduced_rs_manifest.loc[reduced_rs_manifest[\"eol_page_id\"] == pg_id, \"num_images\"] = num_images\n",
+    "\n",
+    "reduced_rs_manifest[\"num_images\"].value_counts()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "One page has only 24, two have 27 & 29, the remaining 395 pages have 30.\n",
+    "\n",
+    "Let's track down those 3 post-it images on page 485420."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "https://eol.org/media/29856158\n",
+      "https://eol.org/media/28605134\n",
+      "https://eol.org/media/14677205\n",
+      "https://eol.org/media/28605135\n",
+      "https://eol.org/media/29856162\n",
+      "https://eol.org/media/29856156\n",
+      "https://eol.org/media/29856167\n",
+      "https://eol.org/media/29856165\n",
+      "https://eol.org/media/29856163\n",
+      "https://eol.org/media/29856157\n",
+      "https://eol.org/media/14677203\n",
+      "https://eol.org/media/29856159\n",
+      "https://eol.org/media/28605131\n",
+      "https://eol.org/media/29856155\n",
+      "https://eol.org/media/29856151\n",
+      "https://eol.org/media/29856161\n",
+      "https://eol.org/media/14677207\n",
+      "https://eol.org/media/29856150\n",
+      "https://eol.org/media/29856164\n",
+      "https://eol.org/media/14677204\n",
+      "https://eol.org/media/28605132\n",
+      "https://eol.org/media/28605133\n",
+      "https://eol.org/media/28605130\n",
+      "https://eol.org/media/29856166\n"
+     ]
+    }
+   ],
+   "source": [
+    "for content_id in list(reduced_rs_manifest.loc[reduced_rs_manifest[\"eol_page_id\"] == 485420, \"eol_content_id\"]):\n",
+    "    print(f\"https://eol.org/media/{content_id}\")\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Post-it image EOL Content IDs (unique in this DataFrame):\n",
+    "- [28605134](https://eol.org/media/28605134)\n",
+    "- [29856159](https://eol.org/media/29856159)\n",
+    "- [29856164](https://eol.org/media/29856164)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "filename                  11983\n",
+       "md5                       11983\n",
+       "combined_id               11983\n",
+       "eol_content_id            11983\n",
+       "eol_page_id                 400\n",
+       "medium_source_url         11983\n",
+       "eol_full_size_copy_url    11983\n",
+       "license_name                 15\n",
+       "copyright_owner            4164\n",
+       "title                      1290\n",
+       "license_link                 10\n",
+       "num_images                    4\n",
+       "keep                          1\n",
+       "dtype: int64"
+      ]
+     },
+     "execution_count": 21,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "post_its = [28605134, 29856159, 29856164]\n",
+    "\n",
+    "rarespecies_manifest = reduced_rs_manifest.loc[~reduced_rs_manifest[\"eol_content_id\"].isin(post_its)]\n",
+    "rarespecies_manifest.nunique()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Remove extra columns"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "Index(['filename', 'md5', 'combined_id', 'eol_content_id', 'eol_page_id',\n",
+       "       'medium_source_url', 'eol_full_size_copy_url', 'license_name',\n",
+       "       'copyright_owner', 'title', 'license_link'],\n",
+       "      dtype='object')"
+      ]
+     },
+     "execution_count": 22,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "rarespecies_manifest = rarespecies_manifest[list(rarespecies_manifest.columns)[:-2]]\n",
+    "rarespecies_manifest.columns"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Add `uuid` (new `rarespecies_id`)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 25,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import uuid"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "11983"
+      ]
+     },
+     "execution_count": 26,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "rarespecies_manifest[\"rarespecies_id\"] = [uuid.uuid4() for i in range(rarespecies_manifest.shape[0])]\n",
+    "rarespecies_manifest.rarespecies_id.nunique()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Save Manifest with `uuid`s"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 27,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "rarespecies_manifest.to_csv(RS_FILEPATH + \"rarespecies_manifest.csv\", index = False)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Make and Save License File"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "licesne_cols = [\"rarespecies_id\",\n",
+    "                \"eol_content_id\",\n",
+    "                \"eol_page_id\",\n",
+    "                \"md5\",\n",
+    "                \"medium_source_url\",\n",
+    "                \"eol_full_size_copy_url\",\n",
+    "                \"license_name\",\n",
+    "                \"copyright_owner\",\n",
+    "                \"license_link\",\n",
+    "                \"title\"]\n",
+    "\n",
+    "licenses = rarespecies_manifest[licesne_cols]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "licenses.to_csv(RS_FILEPATH + \"licenses.csv\", index = False)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Make Rare Species Catalog\n",
+    "\n",
+    "Align taxonomic information to the IDs in the manifest."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 30,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "id_cols = [\"rarespecies_id\", \"eol_content_id\", \"eol_page_id\"]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "['eol_page_id',\n",
+       " 'kingdom',\n",
+       " 'phylum',\n",
+       " 'class',\n",
+       " 'order',\n",
+       " 'family',\n",
+       " 'genus',\n",
+       " 'species',\n",
+       " 'sciName']"
+      ]
+     },
+     "execution_count": 31,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "cat_cols = list(rs_catalog.columns)[2:]\n",
+    "cat_cols"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "rarespecies_id    11983\n",
+       "eol_content_id    11983\n",
+       "eol_page_id         400\n",
+       "kingdom               1\n",
+       "phylum                5\n",
+       "class                15\n",
+       "order                85\n",
+       "family              202\n",
+       "genus               316\n",
+       "species             385\n",
+       "sciName             400\n",
+       "dtype: int64"
+      ]
+     },
+     "execution_count": 32,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "updated_catalog = pd.merge(rarespecies_manifest[id_cols],\n",
+    "                           rs_catalog[cat_cols],\n",
+    "                           on = \"eol_page_id\",\n",
+    "                           how = \"left\")\n",
+    "\n",
+    "updated_catalog.nunique()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Looks good, precisely as expected. Let's save this to a CSV now."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "updated_catalog.to_csv(RS_FILEPATH + \"rarespecies-catalog.csv\", index = False)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "jupytext": {
+   "formats": "ipynb,py:percent"
+  },
+  "kernelspec": {
+   "display_name": "tol",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}