Thresholds.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c73928a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "from category_classification.data_utils import create_dataframe\n",
    "from utils.preprocess import count_categories\n",
    "\n",
    "training_ds = create_dataframe(\"train\", \"xx\")\n",
    "test_ds = create_dataframe(\"test\", \"xx\")\n",
    "val_ds = create_dataframe(\"val\", \"xx\")\n",
    "\n",
    "# Construct the complete dataset from the train/val/test datasets.\n",
    "# complete_ds = training_ds.append(test_ds).append(val_ds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "c0ef2b45",
   "metadata": {},
   "outputs": [],
   "source": [
    "from robotoff.taxonomy import Taxonomy\n",
    "import pandas as pd\n",
    "\n",
    "categories = count_categories(training_ds)\n",
    "category_taxonomy = Taxonomy.from_json(\"data/categories.full.json\")\n",
    "\n",
    "with open(\"data/agribalyse_categories.txt\", \"r\") as f:\n",
    "    argibalyse_categories = set(f.read().split(\",\"))\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "id": "3a3fe5a8",
   "metadata": {},
   "outputs": [],
   "source": [
    "categories_by_filter = pd.DataFrame()\n",
    "\n",
    "categories_by_filter['at least N products'] = range(1, 101)\n",
    "\n",
    "categories_by_filter['Total'] = [[key for key,value in categories.items() if value >= i and key in category_taxonomy] for i in range(1, 101)]\n",
    "categories_by_filter['Agribalyse'] = [[key for key,value in categories.items() if value >= i and key in argibalyse_categories] for i in range(1, 101)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "134ebd8b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "sns.set_style('darkgrid')\n",
    "\n",
    "plt.figure(figsize=(7, 5))\n",
    "plt.ylabel(\"Category Count\")\n",
    "sns.lineplot(y=[len(cat) for cat in categories_by_filter['Total']], x=categories_by_filter['at least N products'], color=\"royalblue\")\n",
    "sns.lineplot(y=[len(cat) for cat in categories_by_filter['Agribalyse']], x=categories_by_filter['at least N products'], color=\"orange\")\n",
    "graph.set_xticks(range(0,110, 10))\n",
    "plt.legend(labels=['All Categories', 'Agribalyse Categories',])\n",
    "\n",
    "plt.show(graph)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "id": "3b3381f0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1e7f36fd0>"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "training_ds['total_category_count'] = [len(x) for x in training_ds['categories_tags']]\n",
    "training_ds['agribalyse_category_count'] = [len([category for category in x if category in argibalyse_categories]) for x in training_ds['categories_tags'] ]\n",
    "\n",
    "graph = sns.histplot([training_ds['total_category_count'], training_ds['agribalyse_category_count']], bins=30)\n",
    "graph.set(xlabel='Categories Per Product', ylabel='Product Count')\n",
    "plt.legend(labels=['Agribalyse Categories','All Categories', ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "id": "a613cfa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_categories_10 = set(categories_by_filter.iloc[9]['Total'])\n",
    "agri_categories_10 = set(categories_by_filter.iloc[9]['Agribalyse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "id": "248a9b9c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3969\n",
      "949\n",
      "6364\n",
      "1747\n"
     ]
    }
   ],
   "source": [
    "print(len(all_categories_10))\n",
    "print(len(agri_categories_10))\n",
    "\n",
    "print(len(categories_by_filter.iloc[0]['Total']))\n",
    "print(len(categories_by_filter.iloc[0]['Agribalyse']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "id": "7db5ff3c",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_ds['total_category_count_10'] = [len([cat for cat in x if cat in all_categories_10]) for x in training_ds['categories_tags']]\n",
    "training_ds['agribalyse_category_count_10'] = [len([category for category in x if category in agri_categories_10]) for x in training_ds['categories_tags'] ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "14894c92",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1e5b55b80>"
      ]
     },
     "execution_count": 200,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "graph = sns.histplot([training_ds['total_category_count_10'], training_ds['agribalyse_category_count_10']], bins=30)\n",
    "graph.set(xlabel='Categories Per Product', ylabel='Product Count')\n",
    "plt.legend(labels=['Agribalyse Categories', 'All Categories'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "id": "721d0595",
   "metadata": {},
   "outputs": [],
   "source": [
    "import argparse\n",
    "import functools\n",
    "import operator\n",
    "import pathlib\n",
    "from typing import Dict\n",
    "\n",
    "from bokeh.plotting import save\n",
    "import pandas as pd\n",
    "from robotoff.taxonomy import Taxonomy\n",
    "from robotoff.utils import gzip_jsonl_iter\n",
    "from tensorflow import keras\n",
    "\n",
    "import settings\n",
    "from category_classification.data_utils import generate_data_from_df\n",
    "from utils.error_analysis import (\n",
    "    generate_analysis_model,\n",
    "    get_deepest_categories,\n",
    "    get_interactive_embedding_plot,\n",
    "    get_error_category,\n",
    ")\n",
    "\n",
    "from utils.io import (\n",
    "    load_config,\n",
    "    load_category_vocabulary,\n",
    "    load_ingredient_vocabulary,\n",
    "    load_product_name_vocabulary,\n",
    ")\n",
    "from utils.metrics import evaluation_report\n",
    "from utils.preprocess import get_nlp\n",
    "\n",
    "\n",
    "\n",
    "model_dir = pathlib.Path(\"weights/0\")\n",
    "model_path =\"weights/0/saved_model\"\n",
    "\n",
    "config = load_config(model_dir)\n",
    "\n",
    "category_to_id = load_category_vocabulary(model_dir)\n",
    "ingredient_to_id = load_ingredient_vocabulary(model_dir)\n",
    "category_names = [\n",
    "    category\n",
    "    for category, _ in sorted(category_to_id.items(), key=operator.itemgetter(1))\n",
    "]\n",
    "\n",
    "nlp = get_nlp(config.lang)\n",
    "\n",
    "product_name_vocabulary = load_product_name_vocabulary(model_dir)\n",
    "model = keras.models.load_model(model_path)\n",
    "\n",
    "generate_data_partial = functools.partial(\n",
    "    generate_data_from_df,\n",
    "    ingredient_to_id=ingredient_to_id,\n",
    "    category_to_id=category_to_id,\n",
    "    product_name_max_length=config.model_config.product_name_max_length,\n",
    "    product_name_token_to_int=product_name_vocabulary,\n",
    "    nlp=nlp,\n",
    "    product_name_preprocessing_config=config.product_name_preprocessing_config,\n",
    "    nutriment_input=config.model_config.nutriment_input,\n",
    ")\n",
    "\n",
    "category_taxonomy: Taxonomy = Taxonomy.from_json(settings.CATEGORY_TAXONOMY_PATH)\n",
    "\n",
    "X_val, y_val = generate_data_partial(test_ds)\n",
    "\n",
    "y_pred_val = model.predict(X_val)\n",
    "\n",
    "predicted = [\n",
    "        [{category_names[i]: conf} for i, conf in sorted(enumerate(y)) if conf >= 0.5]\n",
    "        for y in y_pred_val\n",
    "    ]\n",
    "\n",
    "\n",
    "test_ds[\"predicted categories\"] = [[p for p in preds if next(iter(p)) in categories]\n",
    "    for preds, categories in zip(predicted, test_ds.categories_tags)]\n",
    "\n",
    "\n",
    "test_ds[\"wrong prediction\"] = [[p for p in preds if next(iter(p)) not in categories]\n",
    "        for preds, categories in zip(predicted, test_ds.categories_tags)]\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "id": "87d32426",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = []\n",
    "\n",
    "for index, row in test_ds.iterrows():\n",
    "    lang = row[\"lang\"]\n",
    "    for cat in row[\"predicted categories\"]:\n",
    "        predictions.append((list(cat.values())[0], \"Correct\", lang))\n",
    "    for cat in row[\"wrong prediction\"]:\n",
    "        predictions.append((list(cat.values())[0], \"Incorrect\"))\n",
    "        \n",
    "predictions_df = pd.DataFrame(predictions, columns=[\"Score\", \"Prediction Type\", \"Language\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "id": "2a2f3c33",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "plt.figure(figsize=(10, 7))\n",
    "graph = sns.histplot(data=predictions_df, x='Score', hue='Prediction Type' ,multiple=\"dodge\", bins=10, shrink=.8)\n",
    "graph.set(xlabel='Prediction score', ylabel='Predictions')\n",
    "graph.set_xticks(np.arange(0.5, 1.05, 0.05))\n",
    "plt.gca().invert_xaxis()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "id": "352ce4ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "agribalyse_predictions = []\n",
    "\n",
    "for index, row in test_ds.iterrows():\n",
    "    for cat in row[\"predicted categories\"]:\n",
    "        category = list(cat.values())[0]\n",
    "        if list(cat.keys())[0] not in argibalyse_categories:\n",
    "            continue\n",
    "        agribalyse_predictions.append((list(cat.values())[0], \"Correct\"))\n",
    "    for cat in row[\"wrong prediction\"]:\n",
    "        if list(cat.keys())[0] not in argibalyse_categories:\n",
    "            continue\n",
    "        agribalyse_predictions.append((list(cat.values())[0], \"Incorrect\"))\n",
    "        \n",
    "agribalyse_predictions_df = pd.DataFrame(agribalyse_predictions, columns=[\"Score\", \"Prediction Type\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "id": "137afa81",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 7))\n",
    "graph = sns.histplot(data=agribalyse_predictions_df, x='Score', hue='Prediction Type' ,multiple=\"dodge\", bins=10, shrink=.8, hue_order = ['Correct', 'Incorrect'])\n",
    "graph.set(xlabel='Prediction score', ylabel='Predictions')\n",
    "graph.set_xticks(np.arange(0.5, 1.05, 0.05))\n",
    "plt.gca().invert_xaxis()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "id": "6a6f1253",
   "metadata": {},
   "outputs": [],
   "source": [
    "unique_langs = test_ds.lang.unique()\n",
    "\n",
    "test_ds_by_lang = {elem : pd.DataFrame for elem in unique_langs}\n",
    "\n",
    "for key in test_ds_by_lang.keys():\n",
    "    test_ds_by_lang[key] = test_ds[:][test_ds.lang == key]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "id": "ec450cfc",
   "metadata": {},
   "outputs": [],
   "source": [
    "import operator\n",
    "from typing import Optional, List, Dict\n",
    "\n",
    "import numpy as np\n",
    "from robotoff.taxonomy import Taxonomy\n",
    "\n",
    "from sklearn.metrics import (\n",
    "    precision_score,\n",
    "    recall_score,\n",
    "    f1_score,\n",
    "    classification_report,\n",
    ")\n",
    "\n",
    "\n",
    "def fill_ancestors(\n",
    "    y: np.ndarray,\n",
    "    taxonomy: Taxonomy,\n",
    "    category_to_id: Optional[Dict[str, int]] = None,\n",
    "    category_names: Optional[List[str]] = None,\n",
    "):\n",
    "    if category_to_id is None and category_names is None:\n",
    "        raise ValueError(\"one of category_to_id, category_names must be provided\")\n",
    "\n",
    "    if category_names is None:\n",
    "        category_names = [\n",
    "            cat for cat, _ in sorted(category_to_id.items(), key=operator.itemgetter(1))\n",
    "        ]\n",
    "    elif category_to_id is None:\n",
    "        category_to_id = {cat: i for i, cat in enumerate(category_names)}\n",
    "\n",
    "    y_ = y.copy()\n",
    "    for i in range(y_.shape[1]):\n",
    "        cat_mask = y_[:, i].nonzero()[0]\n",
    "\n",
    "        if len(cat_mask):\n",
    "            category_name = category_names[i]\n",
    "            parents = taxonomy[category_name].get_parents_hierarchy()\n",
    "            parent_ids = [category_to_id[parent.id] for parent in parents]\n",
    "            for parent_id in parent_ids:\n",
    "                y_[cat_mask, parent_id] = 1\n",
    "\n",
    "    return y_\n",
    "\n",
    "\n",
    "def evaluation_report(\n",
    "    y_true: np.ndarray,\n",
    "    y_pred: np.ndarray,\n",
    "    taxonomy: Taxonomy,\n",
    "    category_names: Optional[List[str]],\n",
    "):\n",
    "    y_pred_int = (y_pred > 0.5).astype(y_pred.dtype)\n",
    "    y_pred_int_filled = fill_ancestors(\n",
    "        y_pred_int, taxonomy=taxonomy, category_names=category_names\n",
    "    )\n",
    "\n",
    "    clf_report = None\n",
    "\n",
    "    report = {}\n",
    "\n",
    "    for metric, metric_func in (\n",
    "        (\"precision\", precision_score),\n",
    "        (\"recall\", recall_score),\n",
    "        (\"f1\", f1_score),\n",
    "    ):\n",
    "        for average in (\"micro\", \"macro\"):\n",
    "            metric_value = metric_func(y_true, y_pred_int, average=average)\n",
    "            report[\"{}-{}\".format(average, metric)] = metric_value\n",
    "\n",
    "    return report, clf_report\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "id": "94cdd045",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done with lang: de\n",
      "Done with lang: en\n",
      "Done with lang: es\n",
      "Done with lang: fr\n",
      "Done with lang: it\n",
      "Done with lang: cs\n",
      "Done with lang: pt\n",
      "Done with lang: ru\n",
      "Done with lang: th\n",
      "Done with lang: fi\n",
      "Done with lang: pl\n",
      "Done with lang: el\n",
      "Done with lang: nl\n",
      "Done with lang: ro\n",
      "Done with lang: hu\n",
      "Done with lang: sv\n",
      "Done with lang: sr\n",
      "Done with lang: sq\n",
      "Done with lang: ar\n",
      "Done with lang: nb\n",
      "Done with lang: bs\n",
      "Done with lang: ca\n",
      "Done with lang: hr\n",
      "Done with lang: da\n",
      "Done with lang: mk\n",
      "Done with lang: lb\n",
      "Done with lang: bg\n",
      "Done with lang: tr\n",
      "Done with lang: lt\n",
      "Done with lang: sk\n",
      "Done with lang: sl\n",
      "Done with lang: is\n",
      "Done with lang: la\n",
      "Done with lang: ka\n",
      "Done with lang: fa\n",
      "Done with lang: uk\n",
      "Done with lang: ja\n",
      "Done with lang: he\n",
      "Done with lang: zh\n",
      "Done with lang: lv\n",
      "Done with lang: az\n",
      "Done with lang: id\n",
      "Done with lang: et\n",
      "Done with lang: world\n",
      "Done with lang: no\n",
      "Done with lang: xx\n",
      "Done with lang: ms\n",
      "Done with lang: mr\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/ykulizhskaya/off/robotoff/off-category-classification/venv/lib/python3.8/site-packages/sklearn/metrics/classification.py:1436: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 due to no predicted samples.\n",
      "  precision = _prf_divide(tp_sum, pred_sum,\n",
      "/Users/ykulizhskaya/off/robotoff/off-category-classification/venv/lib/python3.8/site-packages/sklearn/metrics/classification.py:1436: UndefinedMetricWarning: F-score is ill-defined and being set to 0.0 due to no predicted samples.\n",
      "  precision = _prf_divide(tp_sum, pred_sum,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done with lang: jv\n",
      "Done with lang: pt-BR\n",
      "Done with lang: ko\n"
     ]
    }
   ],
   "source": [
    "from category_classification.data_utils import generate_data_from_df\n",
    "\n",
    "\n",
    "lang_metrics = pd.DataFrame(columns=[\"lang\", \"size\", \"micro-precision\", \"micro-recall\"])\n",
    "\n",
    "\n",
    "for key, val in test_ds_by_lang.items():\n",
    "    X, y = generate_data_from_df(\n",
    "        val,\n",
    "        ingredient_to_id,\n",
    "        category_to_id,\n",
    "        product_name_vocabulary,\n",
    "        nlp=nlp,\n",
    "        product_name_max_length=config.model_config.product_name_max_length,\n",
    "        product_name_preprocessing_config=config.product_name_preprocessing_config,\n",
    "        nutriment_input=config.model_config.nutriment_input)\n",
    "    y_pred = model.predict(X)\n",
    "    report, clf_report = evaluation_report(\n",
    "        y, y_pred, taxonomy=category_taxonomy, category_names=category_names\n",
    "    )\n",
    "    lang_metrics = lang_metrics.append({\n",
    "        \"lang\": key,\n",
    "        \"size\":  len(val),\n",
    "        \"micro-precision\": report[\"micro-precision\"],\n",
    "        \"micro-recall\": report[\"micro-recall\"],\n",
    "      }, ignore_index=True)\n",
    "    print(\"Done with lang: {}\".format(key))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "id": "5b9afe8f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(20, 7))\n",
    "graph = sns.barplot(data=lang_metrics[lang_metrics['size'] > 100], x='lang', y='size', palette=\"Paired\")\n",
    "graph.set(xlabel='Lang', ylabel='Number of Samples')\n",
    "sns.set(font_scale = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "id": "09004ace",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='micro-precision', ylabel='micro-recall'>"
      ]
     },
     "execution_count": 283,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8, 6))\n",
    "\n",
    "sns.scatterplot(data=lang_metrics[lang_metrics['size'] > 100], x=\"micro-precision\", y=\"micro-recall\", hue=\"lang\",  palette=\"Paired\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "55d5136a",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}