packages/lib-pacsynth/samples/dp_aggregate_seeded_detailed_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Differential Privacy (DP) Private Aggregate Seeded Synthesizer\n",
    "DP Private Aggregate Seeded synthesizer is a differentially private synthesizer that relies on DP Marginals to build synthetic data. It will compute DP Marginals (called aggregates) for your dataset using a specified `reporting length`, and synthesize data based on the computed aggregated counts.\n",
    "\n",
    "> Aggregates will be computed for all lengths of attribute combination up to and including the `reporting length`.\n",
    "\n",
    "## 1. Overview\n",
    "\n",
    "### 1.1. Aggregate data generation with DP\n",
    "Let's consider the following input as example:\n",
    "\n",
    "| A | B | C |\n",
    "| -- | -- | -- |\n",
    "| a1 | b1 | c1 |\n",
    "| a1 | b2 | c1 |\n",
    "| a2 |    | c2 |\n",
    "| a2 | b2 | c1 |\n",
    "| a1 | b2 |    |\n",
    "\n",
    "The input data is assumed to be categorical and the domain will be inferred from the input dataset:\n",
    "\n",
    "- `A` possible values are `a1,a2`\n",
    "- `B` possible values are `b1,b2`\n",
    "- `C` possible values are `c1,c2`\n",
    "\n",
    "For a `reporting length=2`, the aggregates in the dataset above could be:\n",
    "\n",
    "- 1-counts\n",
    "    - `A:a1`: 3 + NOISE\n",
    "    - `A:a2`: 2 + NOISE\n",
    "    - `B:b1`: 1 + NOISE\n",
    "    - `B:b2`: 3 + NOISE\n",
    "    - `C:c1`: 3 + NOISE\n",
    "    - `C:c2`: 1 + NOISE\n",
    "\n",
    "- 2-counts:\n",
    "    - `A:a1;C:c1`: 2 + NOISE\n",
    "    - `A:a2;B:b2`: 1 + NOISE\n",
    "    - `B:b1;C:c1`: 1 + NOISE\n",
    "    - `A:a1;B:b1`: 1 + NOISE\n",
    "    - `A:a1;B:b2`: 2 + NOISE\n",
    "    - `B:b2;C:c1`: 2 + NOISE\n",
    "    - `A:a2;C:c2`: 1 + NOISE\n",
    "    - `B:b2;C:c2`: 0 + NOISE\n",
    "\n",
    "Also, some spurious combinations might be created and reported to ensure differentially private guarantees - notice that `B:b2;C:c2` does not exist in the sensitive dataset, but it has been _fabricated_ and added to the output.\n",
    "\n",
    "Similarly, some attribute combinations might be suppressed. For example, even though `A:a2;C:c1` exists in the sensitive dataset, it has not been reported as an aggregate.\n",
    "\n",
    "### 1.2. Synthesis\n",
    "Data will be then synthesized directly from the aggregates computed with differential privacy to produce synthetic data. Which will ensure the same DP guarantees to the synthetic data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Imports and global config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import math\n",
    "\n",
    "from pacsynth import init_logger, set_number_of_threads\n",
    "from pacsynth import Dataset\n",
    "from pacsynth import DpAggregateSeededParametersBuilder, AccuracyMode, FabricationMode\n",
    "from pacsynth import DpAggregateSeededSynthesizer, Dataset\n",
    "\n",
    "from utils import gen_data_frame, ErrorReport\n",
    "\n",
    "# The library allows the desired log level to be set if wanted\n",
    "# ('off' || 'error' || 'warn' || 'info' || 'debug' || 'trace')\n",
    "# init_logger('trace')\n",
    "\n",
    "# some algorithms have parallel implementations, so the desired number of threads can be set\n",
    "# (the default is one thread per CPU core)\n",
    "# set_number_of_threads(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Generating an example data frame with random data\n",
    "\n",
    "> `gen_data_frame` is just an utility to generate some example data (the code for it is in [`utils.py`](./utils.py))\n",
    "\n",
    "To illustrate the library, let's start by creating an example data frame:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>H1</th>\n",
       "      <th>H2</th>\n",
       "      <th>H3</th>\n",
       "      <th>H4</th>\n",
       "      <th>H5</th>\n",
       "      <th>H6</th>\n",
       "      <th>H7</th>\n",
       "      <th>H8</th>\n",
       "      <th>H9</th>\n",
       "      <th>H10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5995</th>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5996</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5997</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5998</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5999</th>\n",
       "      <td></td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6000 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     H1 H2  H3 H4 H5 H6 H7 H8 H9 H10\n",
       "0     1  1   4  1  1  0  0  0  0   1\n",
       "1        2   5  0  0  0  0  0  0   1\n",
       "2     2  3   3  0  1  1  0  1  1   0\n",
       "3               0  1  1  0  0  1   1\n",
       "4     2  1      1  1  1  1  0  0   1\n",
       "...  .. ..  .. .. .. .. .. .. ..  ..\n",
       "5995  2  4   7  1  1  0  0  1  0   0\n",
       "5996  1      7  0  1  1  1  0  0   0\n",
       "5997  1         1  1  1  1  1  1   0\n",
       "5998         8  0  0  0  1  1  0   0\n",
       "5999     4  10  0  0  1  1  0  0   0\n",
       "\n",
       "[6000 rows x 10 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "number_of_records_to_generate = 6000\n",
    "\n",
    "sensitive_df = gen_data_frame(number_of_records_to_generate)\n",
    "\n",
    "sensitive_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Creating the sensitive dataset\n",
    "\n",
    "### 4.1. Creation from constructor\n",
    "\n",
    "The library uses an internal representation of the data to optimize execution time complexity.\n",
    "\n",
    "If the data is already in the required raw format, you can call the constructor directly:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>a1</td>\n",
       "      <td>b1</td>\n",
       "      <td>c1</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>a1</td>\n",
       "      <td>b1</td>\n",
       "      <td>c2</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>a2</td>\n",
       "      <td></td>\n",
       "      <td>c2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    A   B   C  D\n",
       "0  a1  b1  c1   \n",
       "1  a1  b1  c2   \n",
       "2  a2      c2  1"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sensitive_raw_data = [\n",
    "    # headers\n",
    "    ['A', 'B', 'C', 'D'],\n",
    "    # records\n",
    "    ['a1', 'b1', 'c1', '0'],\n",
    "    ['a1', 'b1', 'c2', '0'],\n",
    "    ['a2', '', 'c2', '1'],\n",
    "]\n",
    "sensitive_dataset = Dataset(sensitive_raw_data)\n",
    "sensitive_dataset.to_data_frame()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2. Negative value interpretation\n",
    "\n",
    "The library distinguishes 'positive' attribute values that indicate the presence of specific sensitive data from 'negative' attribute values that indicate the absence of such data. By default, the integer zero (`0`) and the empty string (`\"\"`) are not taken into account when creating and counting attribute combinations. Any columns where zero values are of interest (and thus sensitive) should be listed as `sensitive_zeros`, so they will be treated the same way as positive values.\n",
    "\n",
    "> For more parameters see the library documentation - `help('pacsynth.Dataset')`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>a1</td>\n",
       "      <td>b1</td>\n",
       "      <td>c1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>a1</td>\n",
       "      <td>b1</td>\n",
       "      <td>c2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>a2</td>\n",
       "      <td></td>\n",
       "      <td>c2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    A   B   C  D\n",
       "0  a1  b1  c1  0\n",
       "1  a1  b1  c2  0\n",
       "2  a2      c2  1"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sensitive_raw_data = [\n",
    "    # headers\n",
    "    ['A', 'B', 'C', 'D'],\n",
    "    # records\n",
    "    ['a1', 'b1', 'c1', '0'],\n",
    "    ['a1', 'b1', 'c2', '0'],\n",
    "    ['a2', '', 'c2', '1'],\n",
    "]\n",
    "sensitive_dataset = Dataset(sensitive_raw_data, sensitive_zeros=['D'])\n",
    "sensitive_dataset.to_data_frame()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3. Creating sensitive dataset from a pandas data frame\n",
    "For convenience, a method is provided to build a dataset from a pandas data frame:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>H1</th>\n",
       "      <th>H2</th>\n",
       "      <th>H3</th>\n",
       "      <th>H4</th>\n",
       "      <th>H5</th>\n",
       "      <th>H6</th>\n",
       "      <th>H7</th>\n",
       "      <th>H8</th>\n",
       "      <th>H9</th>\n",
       "      <th>H10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5995</th>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5996</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>7</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5997</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5998</th>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>8</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5999</th>\n",
       "      <td></td>\n",
       "      <td>4</td>\n",
       "      <td>10</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6000 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     H1 H2  H3 H4 H5 H6 H7 H8 H9 H10\n",
       "0     1  1   4  1  1               1\n",
       "1        2   5                     1\n",
       "2     2  3   3     1  1     1  1    \n",
       "3                  1  1        1   1\n",
       "4     2  1      1  1  1  1         1\n",
       "...  .. ..  .. .. .. .. .. .. ..  ..\n",
       "5995  2  4   7  1  1        1       \n",
       "5996  1      7     1  1  1          \n",
       "5997  1         1  1  1  1  1  1    \n",
       "5998         8           1  1       \n",
       "5999     4  10        1  1          \n",
       "\n",
       "[6000 rows x 10 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sensitive_dataset = Dataset.from_data_frame(sensitive_df)\n",
    "sensitive_dataset.to_data_frame()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Generating the synthetic data\n",
    "\n",
    "### 5.1. Defining synthesizer parameters\n",
    "\n",
    "If you just want to create the synthesizer with default parameters:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"reporting_length\": 3,\n",
      "  \"epsilon\": 4.0,\n",
      "  \"delta\": null,\n",
      "  \"percentile_percentage\": 99,\n",
      "  \"percentile_epsilon_proportion\": 0.01,\n",
      "  \"sigma_proportions\": [\n",
      "    1.0,\n",
      "    0.5,\n",
      "    0.3333333333333333\n",
      "  ],\n",
      "  \"number_of_records_epsilon_proportion\": 0.005,\n",
      "  \"threshold\": {\n",
      "    \"type\": \"Adaptive\",\n",
      "    \"valuesByLen\": {\n",
      "      \"3\": 1.0,\n",
      "      \"2\": 1.0\n",
      "    }\n",
      "  },\n",
      "  \"empty_value\": \"\",\n",
      "  \"use_synthetic_counts\": false,\n",
      "  \"weight_selection_percentile\": 95,\n",
      "  \"aggregate_counts_scale_factor\": null\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "synth = DpAggregateSeededSynthesizer()\n",
    "print (synth.parameters)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, this might not produce the optimal output for your dataset and downstream analysis tasks.\n",
    "\n",
    "To tune the synthesizer parameters, we provide a builder (`DpAggregateSeededParametersBuilder`). Any parameters passed to the builder will override the default values accordingly:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "  \"reporting_length\": 3,\n",
      "  \"epsilon\": 4.0,\n",
      "  \"delta\": 0.000019158156689251674,\n",
      "  \"percentile_percentage\": 99,\n",
      "  \"percentile_epsilon_proportion\": 0.01,\n",
      "  \"sigma_proportions\": [\n",
      "    1.0,\n",
      "    0.5,\n",
      "    0.3333333333333333\n",
      "  ],\n",
      "  \"number_of_records_epsilon_proportion\": 0.005,\n",
      "  \"threshold\": {\n",
      "    \"type\": \"Adaptive\",\n",
      "    \"valuesByLen\": {\n",
      "      \"3\": 1.0,\n",
      "      \"2\": 1.0\n",
      "    }\n",
      "  },\n",
      "  \"empty_value\": \"\",\n",
      "  \"use_synthetic_counts\": false,\n",
      "  \"weight_selection_percentile\": 95,\n",
      "  \"aggregate_counts_scale_factor\": 1.0\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "params = DpAggregateSeededParametersBuilder() \\\n",
    "        .reporting_length(3) \\\n",
    "        .epsilon(4.0) \\\n",
    "        .delta(1.0 / (math.log(len(sensitive_df)) * len(sensitive_df))) \\\n",
    "        .percentile_percentage(99) \\\n",
    "        .percentile_epsilon_proportion(0.01) \\\n",
    "        .accuracy_mode(AccuracyMode.prioritize_long_combinations()) \\\n",
    "        .number_of_records_epsilon_proportion(0.005) \\\n",
    "        .fabrication_mode(FabricationMode.uncontrolled()) \\\n",
    "        .empty_value(\"\") \\\n",
    "        .weight_selection_percentile(95) \\\n",
    "        .use_synthetic_counts(False) \\\n",
    "        .aggregate_counts_scale_factor(1.0) \\\n",
    "        .build()\n",
    "print (params)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To continue with this example, let's set the parameters we care about for now:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "reporting_length = 4\n",
    "\n",
    "builder = DpAggregateSeededParametersBuilder() \\\n",
    "    .reporting_length(reporting_length) \\\n",
    "    .epsilon(4.0) \\\n",
    "    .accuracy_mode(AccuracyMode.prioritize_long_combinations()) \\\n",
    "    .fabrication_mode(FabricationMode.uncontrolled()) \\\n",
    "    .use_synthetic_counts(True)\n",
    "\n",
    "synth = DpAggregateSeededSynthesizer(builder.build())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2 Building the model and synthesizing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of records protected with DP: 6018\n"
     ]
    },
    {
     "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>H1</th>\n",
       "      <th>H2</th>\n",
       "      <th>H3</th>\n",
       "      <th>H4</th>\n",
       "      <th>H5</th>\n",
       "      <th>H6</th>\n",
       "      <th>H7</th>\n",
       "      <th>H8</th>\n",
       "      <th>H9</th>\n",
       "      <th>H10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td></td>\n",
       "      <td>5</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>5</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6013</th>\n",
       "      <td></td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6014</th>\n",
       "      <td></td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6015</th>\n",
       "      <td></td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6016</th>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>2</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>6017</th>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2</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",
       "  </tbody>\n",
       "</table>\n",
       "<p>6018 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     H1 H2 H3 H4 H5 H6 H7 H8 H9 H10\n",
       "0        5  8  1  1  1  1  1  1   1\n",
       "1        5  8  1  1  1  1  1  1   1\n",
       "2     1     8  1  1  1  1  1  1   1\n",
       "3     1     8  1  1  1  1  1  1   1\n",
       "4     1     8  1  1  1  1  1  1   1\n",
       "...  .. .. .. .. .. .. .. .. ..  ..\n",
       "6013     3  6     1                \n",
       "6014     4  2     1                \n",
       "6015     4  2     1                \n",
       "6016  1  4  2                      \n",
       "6017  2  4  2                      \n",
       "\n",
       "[6018 rows x 10 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "synth.fit(sensitive_dataset)\n",
    "\n",
    "# we could decide to use this or not as the sample number\n",
    "protected_number_of_records = synth.get_dp_number_of_records()\n",
    "\n",
    "print ('Number of records protected with DP:', synth.get_dp_number_of_records())\n",
    "\n",
    "# here if we do not specify the desired number of samples, the synthesizer will\n",
    "# use all the available attributes based on the 1-counts to synthesize records\n",
    "synthetic_raw_data = synth.sample(protected_number_of_records)\n",
    "synthetic_dataset = Dataset(synthetic_raw_data)\n",
    "\n",
    "# as an example, let's create a pandas data frame from the raw synthetic data\n",
    "synthetic_df = Dataset.raw_data_to_data_frame(synthetic_raw_data)\n",
    "synthetic_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Generating/exporting aggregate data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('H1:1;H2:5;H3:3;H4:1', 6),\n",
       " ('H2:6;H3:10;H5:1;H7:1', 16),\n",
       " ('H1:1;H3:6;H4:1;H8:1', 72),\n",
       " ('H1:1;H3:2;H5:1;H8:1', 48),\n",
       " ('H1:1;H3:3;H6:1;H8:1', 14),\n",
       " ('H2:5;H3:9;H6:1;H8:1', 16),\n",
       " ('H1:1;H3:7;H6:1', 97),\n",
       " ('H1:2;H2:5;H3:9;H9:1', 48),\n",
       " ('H2:2;H3:9', 12),\n",
       " ('H2:6;H3:9;H8:1', 71),\n",
       " ('H1:1;H2:1;H3:1;H4:1', 31),\n",
       " ('H10:1;H3:6;H8:1', 99),\n",
       " ('H1:2;H2:5;H4:1', 85),\n",
       " ('H10:1;H2:1;H3:3;H4:1', 35),\n",
       " ('H10:1;H2:1;H3:8', 9),\n",
       " ('H2:4;H3:8;H5:1;H9:1', 34),\n",
       " ('H2:6;H3:2;H7:1;H9:1', 3),\n",
       " ('H10:1;H3:5;H4:1;H6:1', 51),\n",
       " ('H10:1;H1:1;H3:7;H8:1', 23),\n",
       " ('H2:1;H3:2;H6:1;H8:1', 55)]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# generate sensitive aggregates\n",
    "sensitive_aggregates = sensitive_dataset.get_aggregates(reporting_length, ';')\n",
    "\n",
    "# export the differentially private aggregates (internal to the synthesizer)\n",
    "dp_aggregates = synth.get_dp_aggregates(';')\n",
    "\n",
    "# generate aggregates from the synthetic data\n",
    "synthetic_aggregates = synthetic_dataset.get_aggregates(reporting_length, ';')\n",
    "\n",
    "# let's take a look at the DP aggregates\n",
    "list(dp_aggregates.items())[:20]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. Evaluating DP aggregates and DP synthetic data\n",
    "\n",
    "This section is an example evaluation of both the DP aggregates and the DP synthetic data, as well as the influence of some synthesizer parameters in the DP aggregates and synthetic data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.1. Evaluating current results\n",
    "\n",
    "> `ErrorReport` is just an example way to evaluate results (the code for it is [`utils.py`](./utils.py))\n",
    "\n",
    "- **Count**: mean of the aggregate counts for the given length\n",
    "- **Error**: mean of the `abs(sensitive_count - dp_aggregated_count)` or `abs(sensitive_count - synthetic_count)`\n",
    "- **Suppressed %**: percentage of combinations present in the sensitive dataset, but not present in the aggregated/synthetic data\n",
    "- **Fabricated %**: percentage of combinations that were reported in the aggregated/synthetic data, but do not exist in the sensitive dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "sensitive_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in sensitive_aggregates.items()\n",
    "}\n",
    "dp_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in dp_aggregates.items()\n",
    "}\n",
    "synthetic_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in synthetic_aggregates.items()\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Aggregates**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 18.04</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 18.20</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>5.43 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 18.10</td>\n",
       "      <td>0.44 %</td>\n",
       "      <td>6.23 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 13.87</td>\n",
       "      <td>7.42 %</td>\n",
       "      <td>2.99 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 15.31</td>\n",
       "      <td>5.13 %</td>\n",
       "      <td>4.01 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length    Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 18.04       0.00 %       0.00 %\n",
       "1      2   426.81 +/- 18.20       0.00 %       5.43 %\n",
       "2      3   149.56 +/- 18.10       0.44 %       6.23 %\n",
       "3      4    56.41 +/- 13.87       7.42 %       2.99 %\n",
       "4    All   111.70 +/- 15.31       5.13 %       4.01 %"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, dp_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Synthetic Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 81.32</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 64.75</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>5.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 38.39</td>\n",
       "      <td>0.44 %</td>\n",
       "      <td>5.35 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 19.29</td>\n",
       "      <td>7.54 %</td>\n",
       "      <td>2.43 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 27.75</td>\n",
       "      <td>5.22 %</td>\n",
       "      <td>3.38 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length    Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 81.32       0.00 %       0.00 %\n",
       "1      2   426.81 +/- 64.75       0.00 %       5.00 %\n",
       "2      3   149.56 +/- 38.39       0.44 %       5.35 %\n",
       "3      4    56.41 +/- 19.29       7.54 %       2.43 %\n",
       "4    All   111.70 +/- 27.75       5.22 %       3.38 %"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, synthetic_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.2. Targeting less fabrication\n",
    "\n",
    "Let's update the current synthesizer parameters to `minimize` fabrication:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>H1</th>\n",
       "      <th>H2</th>\n",
       "      <th>H3</th>\n",
       "      <th>H4</th>\n",
       "      <th>H5</th>\n",
       "      <th>H6</th>\n",
       "      <th>H7</th>\n",
       "      <th>H8</th>\n",
       "      <th>H9</th>\n",
       "      <th>H10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6013</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>6</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>6014</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>6</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>6015</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>6</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>6016</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>6</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>6017</th>\n",
       "      <td></td>\n",
       "      <td>2</td>\n",
       "      <td>6</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",
       "  </tbody>\n",
       "</table>\n",
       "<p>6018 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     H1 H2 H3 H4 H5 H6 H7 H8 H9 H10\n",
       "0        6     1  1  1  1  1  1   1\n",
       "1        6     1  1  1  1  1  1   1\n",
       "2        6     1  1  1  1  1  1   1\n",
       "3        6     1  1  1  1  1  1   1\n",
       "4        6     1  1  1  1  1  1   1\n",
       "...  .. .. .. .. .. .. .. .. ..  ..\n",
       "6013     2  6                      \n",
       "6014     2  6                      \n",
       "6015     2  6                      \n",
       "6016     2  6                      \n",
       "6017     2  6                      \n",
       "\n",
       "[6018 rows x 10 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "synth = DpAggregateSeededSynthesizer(\n",
    "    builder \\\n",
    "        .fabrication_mode(FabricationMode.minimize()) \\\n",
    "        .build()\n",
    ")\n",
    "\n",
    "synth.fit(sensitive_dataset)\n",
    "\n",
    "synthetic_raw_data = synth.sample(protected_number_of_records)\n",
    "synthetic_dataset = Dataset(synthetic_raw_data)\n",
    "\n",
    "synthetic_df = Dataset.raw_data_to_data_frame(synthetic_raw_data)\n",
    "synthetic_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluating again:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "sensitive_aggregates = sensitive_dataset.get_aggregates(reporting_length, ';')\n",
    "dp_aggregates = synth.get_dp_aggregates(';')\n",
    "synthetic_aggregates = synthetic_dataset.get_aggregates(reporting_length, ';')\n",
    "\n",
    "sensitive_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in sensitive_aggregates.items()\n",
    "}\n",
    "dp_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in dp_aggregates.items()\n",
    "}\n",
    "synthetic_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in synthetic_aggregates.items()\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Aggregates**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 17.76</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 19.90</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.48 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 21.45</td>\n",
       "      <td>20.73 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 17.07</td>\n",
       "      <td>68.37 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 19.27</td>\n",
       "      <td>51.59 %</td>\n",
       "      <td>0.06 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length    Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 17.76       0.00 %       0.00 %\n",
       "1      2   426.81 +/- 19.90       0.00 %       0.48 %\n",
       "2      3   149.56 +/- 21.45      20.73 %       0.00 %\n",
       "3      4    56.41 +/- 17.07      68.37 %       0.00 %\n",
       "4    All   111.70 +/- 19.27      51.59 %       0.06 %"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, dp_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Synthetic Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 274.56</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 169.23</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.48 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 91.08</td>\n",
       "      <td>20.73 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 46.69</td>\n",
       "      <td>68.37 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 83.71</td>\n",
       "      <td>51.59 %</td>\n",
       "      <td>0.06 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length     Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 274.56       0.00 %       0.00 %\n",
       "1      2   426.81 +/- 169.23       0.00 %       0.48 %\n",
       "2      3    149.56 +/- 91.08      20.73 %       0.00 %\n",
       "3      4     56.41 +/- 46.69      68.37 %       0.00 %\n",
       "4    All    111.70 +/- 83.71      51.59 %       0.06 %"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, synthetic_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.3. Prioritize short combinations\n",
    "\n",
    "Let's update the current synthesizer parameters to `prioritize_short_combinations`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "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>H1</th>\n",
       "      <th>H2</th>\n",
       "      <th>H3</th>\n",
       "      <th>H4</th>\n",
       "      <th>H5</th>\n",
       "      <th>H6</th>\n",
       "      <th>H7</th>\n",
       "      <th>H8</th>\n",
       "      <th>H9</th>\n",
       "      <th>H10</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td></td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6013</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td>7</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>6014</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td>7</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>6015</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td>8</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>6016</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td>9</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>6017</th>\n",
       "      <td></td>\n",
       "      <td>6</td>\n",
       "      <td>9</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",
       "  </tbody>\n",
       "</table>\n",
       "<p>6018 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     H1 H2 H3 H4 H5 H6 H7 H8 H9 H10\n",
       "0     2        1  1  1  1  1  1   1\n",
       "1     2        1  1  1  1  1  1   1\n",
       "2     2        1  1  1  1  1  1   1\n",
       "3     2        1  1  1  1  1  1   1\n",
       "4     2        1  1  1  1  1  1   1\n",
       "...  .. .. .. .. .. .. .. .. ..  ..\n",
       "6013     6  7                      \n",
       "6014     6  7                      \n",
       "6015     6  8                      \n",
       "6016     6  9                      \n",
       "6017     6  9                      \n",
       "\n",
       "[6018 rows x 10 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "synth = DpAggregateSeededSynthesizer(\n",
    "    builder \\\n",
    "        .fabrication_mode(FabricationMode.uncontrolled()) \\\n",
    "        .accuracy_mode(AccuracyMode.prioritize_short_combinations()) \\\n",
    "        .use_synthetic_counts(False) \\\n",
    "        .build()\n",
    ")\n",
    "\n",
    "synth.fit(sensitive_dataset)\n",
    "\n",
    "synthetic_raw_data = synth.sample(protected_number_of_records)\n",
    "synthetic_dataset = Dataset(synthetic_raw_data)\n",
    "\n",
    "synthetic_df = Dataset.raw_data_to_data_frame(synthetic_raw_data)\n",
    "synthetic_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Evaluating again:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "sensitive_aggregates = sensitive_dataset.get_aggregates(reporting_length, ';')\n",
    "dp_aggregates = synth.get_dp_aggregates(';')\n",
    "synthetic_aggregates = synthetic_dataset.get_aggregates(reporting_length, ';')\n",
    "\n",
    "sensitive_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in sensitive_aggregates.items()\n",
    "}\n",
    "dp_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in dp_aggregates.items()\n",
    "}\n",
    "synthetic_aggregates_parsed = {\n",
    "    tuple(agg.split(';')): count for (agg, count) in synthetic_aggregates.items()\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Aggregates**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 4.08</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 12.68</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>5.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 25.72</td>\n",
       "      <td>1.76 %</td>\n",
       "      <td>4.91 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 29.15</td>\n",
       "      <td>32.59 %</td>\n",
       "      <td>3.01 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 26.54</td>\n",
       "      <td>22.51 %</td>\n",
       "      <td>3.77 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length   Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 4.08       0.00 %       0.00 %\n",
       "1      2  426.81 +/- 12.68       0.00 %       5.00 %\n",
       "2      3  149.56 +/- 25.72       1.76 %       4.91 %\n",
       "3      4   56.41 +/- 29.15      32.59 %       3.01 %\n",
       "4    All  111.70 +/- 26.54      22.51 %       3.77 %"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, dp_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Sensitive Data vs. DP Synthetic Data**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>Length</th>\n",
       "      <th>Count +/- Error</th>\n",
       "      <th>Suppressed %</th>\n",
       "      <th>Fabricated %</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1383.32 +/- 84.68</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>0.00 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>426.81 +/- 109.37</td>\n",
       "      <td>0.00 %</td>\n",
       "      <td>4.13 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>149.56 +/- 72.41</td>\n",
       "      <td>2.09 %</td>\n",
       "      <td>3.37 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>56.41 +/- 43.12</td>\n",
       "      <td>34.39 %</td>\n",
       "      <td>1.94 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>All</td>\n",
       "      <td>111.70 +/- 58.33</td>\n",
       "      <td>23.81 %</td>\n",
       "      <td>2.57 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Length    Count +/- Error Suppressed % Fabricated %\n",
       "0      1  1383.32 +/- 84.68       0.00 %       0.00 %\n",
       "1      2  426.81 +/- 109.37       0.00 %       4.13 %\n",
       "2      3   149.56 +/- 72.41       2.09 %       3.37 %\n",
       "3      4    56.41 +/- 43.12      34.39 %       1.94 %\n",
       "4    All   111.70 +/- 58.33      23.81 %       2.57 %"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ErrorReport(sensitive_aggregates_parsed, synthetic_aggregates_parsed).gen()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8. Library documentation\n",
    "\n",
    "The library is fully documented, to check the documentation you can:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# you can check the docs for the entire library\n",
    "# help('pacsynth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on class DpAggregateSeededSynthesizer in pacsynth:\n",
      "\n",
      "pacsynth.DpAggregateSeededSynthesizer = class DpAggregateSeededSynthesizer(object)\n",
      " |  pacsynth.DpAggregateSeededSynthesizer(parameters=None)\n",
      " |  \n",
      " |  Differential Privacy (DP) Private Aggregate Seeded Synthesizer.\n",
      " |  \n",
      " |  DP Private Aggregate Seeded synthesizer is a differentially private synthesizer that relies on\n",
      " |  DP Marginals to build synthetic data. It will compute DP Marginals (called aggregates)\n",
      " |  for your dataset (`.fit`) using the specified parameters, and synthesize data (`.sample`) based on the\n",
      " |  computed aggregated counts (`.get_dp_aggregates`).\n",
      " |  \n",
      " |  Arguments:\n",
      " |      * parameters: Optional[DpAggregateSeededParameters] - parameters constructed with DpAggregateSeededParametersBuilder\n",
      " |          - if not provided, default parameters will be used: `DpAggregateSeededParametersBuilder().build()`\n",
      " |  \n",
      " |  Returns:\n",
      " |      New DpAggregateSeededSynthesizer\n",
      " |  \n",
      " |  Methods defined here:\n",
      " |  \n",
      " |  fit(self)\n",
      " |      Computes the differentially private aggregates used to synthesize\n",
      " |      data.\n",
      " |      \n",
      " |      Arguments:\n",
      " |          * dataset: Dataset - dataset to compute the aggregates for\n",
      " |  \n",
      " |  get_dp_aggregates(self, combination_delimiter)\n",
      " |      Gets the differentially private aggregates computed with the `.fit` method.\n",
      " |      \n",
      " |      Arguments:\n",
      " |          * combination_delimiter: str - combination delimiter to use\n",
      " |      \n",
      " |      Returns:\n",
      " |          A dictionary with the combination string representation as key\n",
      " |          and the combination count as value - dict[str, int]\n",
      " |  \n",
      " |  get_dp_number_of_records(self)\n",
      " |      Gets the differentially private number of records computed with the `.fit` method.\n",
      " |      \n",
      " |      This is different than the number of records specified in the sample method, synthesized\n",
      " |      in the synthetic data. This refers to the differentially private protected number of records\n",
      " |      in original sensitive dataset.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Protected number of records - int\n",
      " |  \n",
      " |  sample(self, target_number_of_records=None, join_multi_value_columns=None)\n",
      " |      Sample records from the differentially private aggregates.\n",
      " |      \n",
      " |      Arguments:\n",
      " |          * target_number_of_records: Optional[int] - desired number of records to be synthesized\n",
      " |              - if not set, the synthesizer will use all the available differentially private attributes\n",
      " |                counts to synthesize records (which will produce a number close to original number of records)\n",
      " |          * join_multi_value_columns: Optional[bool] - whether previous spread multi value columns in the input\n",
      " |                                                       dataset should be joined back together in the resulting raw data\n",
      " |      \n",
      " |      Returns:\n",
      " |          The result raw data form the synthesis,\n",
      " |          where the first entry contains the headers - list[list[str]]\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Static methods defined here:\n",
      " |  \n",
      " |  __new__(*args, **kwargs) from builtins.type\n",
      " |      Create and return a new object.  See help(type) for accurate signature.\n",
      " |  \n",
      " |  ----------------------------------------------------------------------\n",
      " |  Data descriptors defined here:\n",
      " |  \n",
      " |  parameters\n",
      " |      Returns the configured parameters for the synthesizer.\n",
      " |      \n",
      " |      Returns:\n",
      " |          Configured parameters - DpAggregateSeededParameters\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# for a class\n",
    "help('pacsynth.DpAggregateSeededSynthesizer')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on built-in function prioritize_long_combinations in pacsynth.AccuracyMode:\n",
      "\n",
      "pacsynth.AccuracyMode.prioritize_long_combinations = prioritize_long_combinations()\n",
      "    This mode will ensure that more privacy budget is spent for\n",
      "    for longer attribute combination lengths.\n",
      "    \n",
      "    For example, if reporting_length=3 and S(i) the scale of a gaussian noise\n",
      "    added to the correspondent combination length:\n",
      "        - single attribute counts      (1-counts) = S(1)\n",
      "        - combinations of 2 attributes (2-counts) = S(2) = S(1) / 2\n",
      "        - combinations of 3 attributes (3-counts) = S(3) = S(1) / 3\n",
      "    \n",
      "    So 3 times MORE BUDGET is going to be spent with the 3-counts\n",
      "    than with the 1-counts, meaning that the scale of noise related to the 1-counts\n",
      "    will be 3 times bigger than the scale related with the 3-counts.\n",
      "    \n",
      "    Summary:\n",
      "        Use this if you want smaller errors for longer attribute combination lengths\n",
      "        (e.g. the accuracy for 3-counts is more important than for 1-counts)\n",
      "    \n",
      "    Returns:\n",
      "        AccuracyMode\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# for a method\n",
    "help('pacsynth.AccuracyMode.prioritize_long_combinations')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on built-in function init_logger in pacsynth:\n",
      "\n",
      "pacsynth.init_logger = init_logger(level_str)\n",
      "    Enables logging and sets the desired log level.\n",
      "    This is supposed to be called only once.\n",
      "    \n",
      "    Arguments:\n",
      "        * level_str: str - 'off' || 'error' || 'warn' || 'info' || 'debug' || 'trace'\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# for a function\n",
    "help('pacsynth.init_logger')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.4 ('.sds-env': venv)",
   "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": "3.10.6"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "6e516db57ad967ab494bdf699f413a90667b7090ebbf1a23d1813c138c5cb21f"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}