{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Working with tabular data: Pandas\n",
    "\n",
    "Pandas provides three things: \n",
    "1. a new data type specifically designed for tabular data - the `DataFrame`\n",
    "2. functions for manipulating tabular data\n",
    "3. IO for tabular data (covered in Week3/Files)\n",
    "\n",
    "## Why yet another data type?\n",
    "Numpy arrays can hold tabular data - 2D matrices. So why do we need another, special data type?\n",
    "One problem with raw 2D arrays is that they are not self-documenting. What does that mean?\n",
    "\n",
    "Take the data from exercise of Week3/Files: We calculated behavioral scores, stored them in a numpy array, and saved the array to a text file:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.08656877, 0.26186048],\n",
       "       [0.27081788, 0.48948194],\n",
       "       [0.47072199, 0.65700287],\n",
       "       [0.55996639, 0.81680243],\n",
       "       [0.84880959, 0.98686545],\n",
       "       [0.91045131, 0.99465925]])"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "# load and print matrix\n",
    "scores = np.loadtxt('dat/pd_scores.txt')\n",
    "scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Do you remember what the data means? What are the individual rows? What are the columns? Imagine this were an analysis on your own data and you'd want to look at the results in 3 months or so.\n",
    "\n",
    "There are no labels, so it's hard to know from the data itself what they mean - the data is not \"self-documenting\"\n",
    "\n",
    "We can turn this into a DataFrame:"
   ]
  },
  {
   "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>intial score</th>\n",
       "      <th>final score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>20230912</th>\n",
       "      <td>0.086569</td>\n",
       "      <td>0.261860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230913</th>\n",
       "      <td>0.270818</td>\n",
       "      <td>0.489482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230914</th>\n",
       "      <td>0.470722</td>\n",
       "      <td>0.657003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230915</th>\n",
       "      <td>0.559966</td>\n",
       "      <td>0.816802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230916</th>\n",
       "      <td>0.848810</td>\n",
       "      <td>0.986865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230917</th>\n",
       "      <td>0.910451</td>\n",
       "      <td>0.994659</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          intial score  final score\n",
       "20230912      0.086569     0.261860\n",
       "20230913      0.270818     0.489482\n",
       "20230914      0.470722     0.657003\n",
       "20230915      0.559966     0.816802\n",
       "20230916      0.848810     0.986865\n",
       "20230917      0.910451     0.994659"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "# the folder names correspond to the days\n",
    "days = ['20230912','20230913','20230914','20230915','20230916','20230917']\n",
    "\n",
    "# make DataFrame - no need to understand the specifics - I will explain soon what is going on here\n",
    "# we can easily turn a dictionary into a DataFrame - keys will be column labels, values are the per-row data\n",
    "# we can also specify row labels - an index.\n",
    "scores_dict = {'intial score': scores[:, 0], 'final score': scores[:, 1]}\n",
    "df = pd.DataFrame(scores_dict, index=days)\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can save the data frame as csv and excel, and reload them as a DataFrame, with the labels being preserved:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>Unnamed: 0</th>\n",
       "      <th>final score</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>intial score</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.086569</th>\n",
       "      <td>20230912</td>\n",
       "      <td>0.261860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.270818</th>\n",
       "      <td>20230913</td>\n",
       "      <td>0.489482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.470722</th>\n",
       "      <td>20230914</td>\n",
       "      <td>0.657003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.559966</th>\n",
       "      <td>20230915</td>\n",
       "      <td>0.816802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.848810</th>\n",
       "      <td>20230916</td>\n",
       "      <td>0.986865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.910451</th>\n",
       "      <td>20230917</td>\n",
       "      <td>0.994659</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              Unnamed: 0  final score\n",
       "intial score                         \n",
       "0.086569        20230912     0.261860\n",
       "0.270818        20230913     0.489482\n",
       "0.470722        20230914     0.657003\n",
       "0.559966        20230915     0.816802\n",
       "0.848810        20230916     0.986865\n",
       "0.910451        20230917     0.994659"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.to_csv('dat/scores_df.csv')\n",
    "df.to_excel('dat/scores_df.xlsx')\n",
    "\n",
    "# load data\n",
    "df_from_file = pd.read_csv('dat/scores_df.csv', index_col=0, )\n",
    "df_from_file"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Creating DataFrames\n",
    "\n",
    "We already learned that pandas provides versatile file I/O (csv, excel): [https://pandas.pydata.org/docs/user_guide/io.html]() (see Week3/Files)\n",
    "\n",
    "As shown above, a DataFrame can be created from Dictionary.\n",
    "\n",
    "Data is 2D and is organized in columns and rows (rows=index). If we do not specify an index during DataFrame creation, it will be generated automatically, as a row numbers:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'column 1': ['a', 'b', 'c'], 'column 2': [10, 20, 30]}\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>column 1</th>\n",
       "      <th>column 2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>a</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>b</td>\n",
       "      <td>20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>c</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  column 1  column 2\n",
       "0        a        10\n",
       "1        b        20\n",
       "2        c        30"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d = {'column 1': ['a','b','c'], 'column 2': [10, 20, 30]}\n",
    "\n",
    "print(d)\n",
    "df = pd.DataFrame(d)\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A DataFrame can also be created from a 2D np.array. In that case, we can specify the column names separately:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['20230912', '20230913', '20230914', '20230915', '20230916', '20230917']"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# scores\n",
    "days"
   ]
  },
  {
   "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>initial</th>\n",
       "      <th>final</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>20230912</th>\n",
       "      <td>0.086569</td>\n",
       "      <td>0.261860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230913</th>\n",
       "      <td>0.270818</td>\n",
       "      <td>0.489482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230914</th>\n",
       "      <td>0.470722</td>\n",
       "      <td>0.657003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230915</th>\n",
       "      <td>0.559966</td>\n",
       "      <td>0.816802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230916</th>\n",
       "      <td>0.848810</td>\n",
       "      <td>0.986865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230917</th>\n",
       "      <td>0.910451</td>\n",
       "      <td>0.994659</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           initial     final\n",
       "20230912  0.086569  0.261860\n",
       "20230913  0.270818  0.489482\n",
       "20230914  0.470722  0.657003\n",
       "20230915  0.559966  0.816802\n",
       "20230916  0.848810  0.986865\n",
       "20230917  0.910451  0.994659"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(scores, columns=['initial', 'final'], index=days)\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Accessing data in DataFrames\n",
    "There are many ways of accessing data in a DataFrame. We cover the basics here.\n",
    "\n",
    "More details: [https://pandas.pydata.org/docs/user_guide/indexing.html]()\n",
    "\n",
    "### Access columns like a dictionary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20230912    0.261860\n",
       "20230913    0.489482\n",
       "20230914    0.657003\n",
       "20230915    0.816802\n",
       "20230916    0.986865\n",
       "20230917    0.994659\n",
       "Name: final, dtype: float64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['final']  # df[column_name]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Access rows and columns by name like a 2D dictionary via df.loc\n",
    "\n",
    "`df.loc[index_name, column_name]`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>initial</th>\n",
       "      <th>final</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>20230912</th>\n",
       "      <td>0.086569</td>\n",
       "      <td>0.261860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230913</th>\n",
       "      <td>0.270818</td>\n",
       "      <td>0.489482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230914</th>\n",
       "      <td>0.470722</td>\n",
       "      <td>0.657003</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230915</th>\n",
       "      <td>0.559966</td>\n",
       "      <td>0.816802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230916</th>\n",
       "      <td>0.848810</td>\n",
       "      <td>0.986865</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230917</th>\n",
       "      <td>0.910451</td>\n",
       "      <td>0.994659</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           initial     final\n",
       "20230912  0.086569  0.261860\n",
       "20230913  0.270818  0.489482\n",
       "20230914  0.470722  0.657003\n",
       "20230915  0.559966  0.816802\n",
       "20230916  0.848810  0.986865\n",
       "20230917  0.910451  0.994659"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "20230912    0.086569\n",
       "20230913    0.270818\n",
       "20230914    0.470722\n",
       "20230915    0.559966\n",
       "20230916    0.848810\n",
       "20230917    0.910451\n",
       "Name: initial, dtype: float64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(scores, columns=['initial', 'final'], index=days)\n",
    "display(df)\n",
    "\n",
    "df.loc[:, 'initial']\n",
    "\n",
    "\n",
    "# print(\"by row name - all columns: df.loc['20230915']\")\n",
    "# print(df.loc['20230915'])\n",
    "\n",
    "# print(\"this is equivalent to: df.loc['20230915', :]\")\n",
    "# print(df.loc['20230915', :])\n",
    "\n",
    "# print(\"by column name - all rows, one column: df.loc[:, 'initial']\")\n",
    "# print(df.loc[:, 'initial'])\n",
    "\n",
    "# print(\"\\nby row and column name - returns a single cell in the table: df.loc['20230915', 'initial']\")\n",
    "# print(df.loc['20230915', 'initial'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Access to rows and columns via a numerical index like a 2D array via df.iloc\n",
    "\n",
    "`df.iloc[row_number, column_number]`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20230912    0.261860\n",
      "20230913    0.489482\n",
      "20230914    0.657003\n",
      "20230915    0.816802\n",
      "20230916    0.986865\n",
      "20230917    0.994659\n",
      "Name: final, dtype: float64\n",
      "initial    0.270818\n",
      "final      0.489482\n",
      "Name: 20230913, dtype: float64\n",
      "initial    0.270818\n",
      "final      0.489482\n",
      "Name: 20230913, dtype: float64\n",
      "0.6570028706902653\n",
      "20230912    0.261860\n",
      "20230913    0.489482\n",
      "Name: final, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(df.iloc[:, 1])\n",
    "print(df.iloc[1])\n",
    "print(df.iloc[1, :])\n",
    "print(df.iloc[2, 1])\n",
    "print(df.iloc[:2, 1])  # slicing also works"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Mini-Exercise:\n",
    "Get from the DataFrame below:\n",
    "- all data from the 'test' column\n",
    "- all data from the '...' row\n",
    "- data from the cell at row '...' and column '...'\n",
    "- the data from the 5th row"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame(...)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Access the underlying data like a numpy array:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[0.08656877, 0.26186048],\n",
       "        [0.27081788, 0.48948194],\n",
       "        [0.47072199, 0.65700287],\n",
       "        [0.55996639, 0.81680243],\n",
       "        [0.84880959, 0.98686545],\n",
       "        [0.91045131, 0.99465925]]),\n",
       " numpy.ndarray)"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = df.values\n",
    "data, type(data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Boolean indexing also works"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>initial</th>\n",
       "      <th>final</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>20230912</th>\n",
       "      <td>0.086569</td>\n",
       "      <td>0.261860</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20230913</th>\n",
       "      <td>0.270818</td>\n",
       "      <td>0.489482</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           initial     final\n",
       "20230912  0.086569  0.261860\n",
       "20230913  0.270818  0.489482"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['initial'] < 0.4]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "You can apply boolean indices repeatedly, to filter by different conditions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "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>name</th>\n",
       "      <th>age</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Tim</td>\n",
       "      <td>12</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Jim</td>\n",
       "      <td>24</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Pim</td>\n",
       "      <td>18</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Pip</td>\n",
       "      <td>22</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Tom</td>\n",
       "      <td>26</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  name  age  score\n",
       "0  Tim   12      4\n",
       "1  Jim   24      0\n",
       "2  Pim   18      8\n",
       "3  Pip   22      9\n",
       "4  Tom   26      7"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>name</th>\n",
       "      <th>age</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Jim</td>\n",
       "      <td>24</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Pip</td>\n",
       "      <td>22</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Tom</td>\n",
       "      <td>26</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  name  age  score\n",
       "1  Jim   24      0\n",
       "3  Pip   22      9\n",
       "4  Tom   26      7"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>name</th>\n",
       "      <th>age</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Pip</td>\n",
       "      <td>22</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Tom</td>\n",
       "      <td>26</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  name  age  score\n",
       "3  Pip   22      9\n",
       "4  Tom   26      7"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = pd.DataFrame({'name': ['Tim', 'Jim', 'Pim', 'Pip', 'Tom'],\n",
    "                   'age': [12, 24, 18, 22, 26],\n",
    "                   'score': [4, 0, 8, 9, 7]})\n",
    "display(df)\n",
    "\n",
    "# first, keep only the old one - age>20 yrs\n",
    "old_guys = df[df['age']> 20]\n",
    "display(old_guys)\n",
    "\n",
    "# then, filter the dataframe with the old ones, to keep only the high scores - score>4\n",
    "high_scorer = old_guys[old_guys['score']> 4]\n",
    "display(high_scorer)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Groupby: Split-apply-combine\n",
    "A common problem in data analysis is:\n",
    "\n",
    "We have an experiment with multiple subjects, and multiple measurements per subject. We'd like to compute the average score for each subject for plotting and statistics.\n",
    "\n",
    "For instance:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# REMOVE - make dummy data for grouped stats\n",
    "import matplotlib.pyplot as plt\n",
    "nb_animals = 10\n",
    "nb_trials = 6\n",
    "\n",
    "scores = np.zeros((nb_animals * nb_trials, 3))\n",
    "for ani, animal in enumerate(range(nb_animals)):\n",
    "    scores[(ani) * nb_trials:(ani + 1) * nb_trials, 1:] = 2+np.random.randn(nb_trials, 2)/2 + np.array([0.3, 0.8]) + np.random.randn(1) / 2\n",
    "    scores[(ani) * nb_trials:(ani + 1) * nb_trials, 0] = animal\n",
    "\n",
    "plt.plot(scores)\n",
    "\n",
    "df = pd.DataFrame({'animal': scores[:, 0], 'pre': scores[:,1], 'post': scores[:,2]})\n",
    "df.to_csv('dat/groupby.csv', index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "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>animal</th>\n",
       "      <th>pre</th>\n",
       "      <th>post</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.211986</td>\n",
       "      <td>3.107796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.204871</td>\n",
       "      <td>2.830730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.560633</td>\n",
       "      <td>3.483980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.476412</td>\n",
       "      <td>2.801709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>3.662519</td>\n",
       "      <td>3.447005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.538696</td>\n",
       "      <td>3.301515</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.867229</td>\n",
       "      <td>3.484270</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1.0</td>\n",
       "      <td>3.306150</td>\n",
       "      <td>3.077195</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2.741388</td>\n",
       "      <td>2.586878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.874149</td>\n",
       "      <td>3.264328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1.0</td>\n",
       "      <td>3.276006</td>\n",
       "      <td>2.503612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1.0</td>\n",
       "      <td>3.363795</td>\n",
       "      <td>3.175878</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>2.0</td>\n",
       "      <td>3.645122</td>\n",
       "      <td>3.871356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>2.0</td>\n",
       "      <td>4.419345</td>\n",
       "      <td>3.223414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2.0</td>\n",
       "      <td>3.157167</td>\n",
       "      <td>2.851897</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>2.0</td>\n",
       "      <td>3.075222</td>\n",
       "      <td>3.591049</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>2.0</td>\n",
       "      <td>3.085689</td>\n",
       "      <td>4.618074</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2.799818</td>\n",
       "      <td>3.586372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>3.0</td>\n",
       "      <td>2.717420</td>\n",
       "      <td>3.483265</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1.267145</td>\n",
       "      <td>2.402146</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3.497071</td>\n",
       "      <td>2.818079</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1.207408</td>\n",
       "      <td>2.662157</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>3.0</td>\n",
       "      <td>2.392043</td>\n",
       "      <td>2.872839</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>3.0</td>\n",
       "      <td>2.845028</td>\n",
       "      <td>2.683747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.963877</td>\n",
       "      <td>1.633267</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>4.0</td>\n",
       "      <td>2.557404</td>\n",
       "      <td>1.549594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.115516</td>\n",
       "      <td>2.964476</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>4.0</td>\n",
       "      <td>1.830395</td>\n",
       "      <td>1.952609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>4.0</td>\n",
       "      <td>2.170095</td>\n",
       "      <td>2.525734</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>4.0</td>\n",
       "      <td>2.586667</td>\n",
       "      <td>2.240448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>5.0</td>\n",
       "      <td>2.045121</td>\n",
       "      <td>3.573011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>5.0</td>\n",
       "      <td>2.054961</td>\n",
       "      <td>3.010360</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>5.0</td>\n",
       "      <td>2.021420</td>\n",
       "      <td>3.538843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>5.0</td>\n",
       "      <td>1.626790</td>\n",
       "      <td>2.702867</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>5.0</td>\n",
       "      <td>2.342304</td>\n",
       "      <td>2.389883</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>5.0</td>\n",
       "      <td>1.800452</td>\n",
       "      <td>3.311980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>6.0</td>\n",
       "      <td>0.641648</td>\n",
       "      <td>1.322786</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>6.0</td>\n",
       "      <td>2.275206</td>\n",
       "      <td>0.707293</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>6.0</td>\n",
       "      <td>1.090543</td>\n",
       "      <td>1.291870</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>6.0</td>\n",
       "      <td>1.289345</td>\n",
       "      <td>1.976258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>6.0</td>\n",
       "      <td>1.828772</td>\n",
       "      <td>1.341221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>6.0</td>\n",
       "      <td>0.982321</td>\n",
       "      <td>2.676962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>7.0</td>\n",
       "      <td>3.179301</td>\n",
       "      <td>4.185368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>7.0</td>\n",
       "      <td>3.756986</td>\n",
       "      <td>3.681570</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>7.0</td>\n",
       "      <td>2.070850</td>\n",
       "      <td>3.249996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>7.0</td>\n",
       "      <td>2.405634</td>\n",
       "      <td>3.025023</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>7.0</td>\n",
       "      <td>3.773297</td>\n",
       "      <td>3.787543</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>7.0</td>\n",
       "      <td>3.351603</td>\n",
       "      <td>3.171412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>8.0</td>\n",
       "      <td>2.075409</td>\n",
       "      <td>3.820043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>8.0</td>\n",
       "      <td>1.989337</td>\n",
       "      <td>2.970652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>8.0</td>\n",
       "      <td>2.469318</td>\n",
       "      <td>2.046087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>8.0</td>\n",
       "      <td>2.054879</td>\n",
       "      <td>2.514335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>8.0</td>\n",
       "      <td>1.448087</td>\n",
       "      <td>2.380560</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>8.0</td>\n",
       "      <td>2.552318</td>\n",
       "      <td>3.016450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>9.0</td>\n",
       "      <td>2.369584</td>\n",
       "      <td>1.889417</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>9.0</td>\n",
       "      <td>3.086123</td>\n",
       "      <td>2.560759</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>9.0</td>\n",
       "      <td>2.097187</td>\n",
       "      <td>2.446292</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>9.0</td>\n",
       "      <td>1.857415</td>\n",
       "      <td>0.800996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>9.0</td>\n",
       "      <td>1.498517</td>\n",
       "      <td>1.811201</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>9.0</td>\n",
       "      <td>0.719002</td>\n",
       "      <td>2.691326</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    animal       pre      post\n",
       "0      0.0  2.211986  3.107796\n",
       "1      0.0  2.204871  2.830730\n",
       "2      0.0  2.560633  3.483980\n",
       "3      0.0  1.476412  2.801709\n",
       "4      0.0  3.662519  3.447005\n",
       "5      0.0  2.538696  3.301515\n",
       "6      1.0  1.867229  3.484270\n",
       "7      1.0  3.306150  3.077195\n",
       "8      1.0  2.741388  2.586878\n",
       "9      1.0  1.874149  3.264328\n",
       "10     1.0  3.276006  2.503612\n",
       "11     1.0  3.363795  3.175878\n",
       "12     2.0  3.645122  3.871356\n",
       "13     2.0  4.419345  3.223414\n",
       "14     2.0  3.157167  2.851897\n",
       "15     2.0  3.075222  3.591049\n",
       "16     2.0  3.085689  4.618074\n",
       "17     2.0  2.799818  3.586372\n",
       "18     3.0  2.717420  3.483265\n",
       "19     3.0  1.267145  2.402146\n",
       "20     3.0  3.497071  2.818079\n",
       "21     3.0  1.207408  2.662157\n",
       "22     3.0  2.392043  2.872839\n",
       "23     3.0  2.845028  2.683747\n",
       "24     4.0  1.963877  1.633267\n",
       "25     4.0  2.557404  1.549594\n",
       "26     4.0  1.115516  2.964476\n",
       "27     4.0  1.830395  1.952609\n",
       "28     4.0  2.170095  2.525734\n",
       "29     4.0  2.586667  2.240448\n",
       "30     5.0  2.045121  3.573011\n",
       "31     5.0  2.054961  3.010360\n",
       "32     5.0  2.021420  3.538843\n",
       "33     5.0  1.626790  2.702867\n",
       "34     5.0  2.342304  2.389883\n",
       "35     5.0  1.800452  3.311980\n",
       "36     6.0  0.641648  1.322786\n",
       "37     6.0  2.275206  0.707293\n",
       "38     6.0  1.090543  1.291870\n",
       "39     6.0  1.289345  1.976258\n",
       "40     6.0  1.828772  1.341221\n",
       "41     6.0  0.982321  2.676962\n",
       "42     7.0  3.179301  4.185368\n",
       "43     7.0  3.756986  3.681570\n",
       "44     7.0  2.070850  3.249996\n",
       "45     7.0  2.405634  3.025023\n",
       "46     7.0  3.773297  3.787543\n",
       "47     7.0  3.351603  3.171412\n",
       "48     8.0  2.075409  3.820043\n",
       "49     8.0  1.989337  2.970652\n",
       "50     8.0  2.469318  2.046087\n",
       "51     8.0  2.054879  2.514335\n",
       "52     8.0  1.448087  2.380560\n",
       "53     8.0  2.552318  3.016450\n",
       "54     9.0  2.369584  1.889417\n",
       "55     9.0  3.086123  2.560759\n",
       "56     9.0  2.097187  2.446292\n",
       "57     9.0  1.857415  0.800996\n",
       "58     9.0  1.498517  1.811201\n",
       "59     9.0  0.719002  2.691326"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('dat/pd_groupby.csv')\n",
    "df\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "We can solve this using for loops and boolean indexing:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "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>animal</th>\n",
       "      <th>pre</th>\n",
       "      <th>post</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.211986</td>\n",
       "      <td>3.107796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.204871</td>\n",
       "      <td>2.830730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.560633</td>\n",
       "      <td>3.483980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.476412</td>\n",
       "      <td>2.801709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>3.662519</td>\n",
       "      <td>3.447005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.538696</td>\n",
       "      <td>3.301515</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   animal       pre      post\n",
       "0     0.0  2.211986  3.107796\n",
       "1     0.0  2.204871  2.830730\n",
       "2     0.0  2.560633  3.483980\n",
       "3     0.0  1.476412  2.801709\n",
       "4     0.0  3.662519  3.447005\n",
       "5     0.0  2.538696  3.301515"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "animal    0.000000\n",
       "pre       2.442519\n",
       "post      3.162122\n",
       "dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "animal    0.000000\n",
       "pre       2.442519\n",
       "post      3.162122\n",
       "dtype: float64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "[array([0.        , 2.44251943, 3.16212242]),\n",
       " array([1.        , 2.73811958, 3.01536034]),\n",
       " array([2.        , 3.36372726, 3.62369364]),\n",
       " array([3.        , 2.32101921, 2.82037202]),\n",
       " array([4.        , 2.03732544, 2.14435453]),\n",
       " array([5.        , 1.98184125, 3.0878241 ]),\n",
       " array([6.        , 1.35130571, 1.55273191]),\n",
       " array([7.        , 3.08961177, 3.51681884]),\n",
       " array([8.        , 2.09822498, 2.79135462]),\n",
       " array([9.        , 1.9379714 , 2.03333186])]"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# identify all animals\n",
    "animals = df['animal'].unique()\n",
    "\n",
    "# example of boolean indexing\n",
    "display(df[df['animal']==0.0])\n",
    "# compute the mean:\n",
    "display(df[df['animal']==0.0].mean())\n",
    "display(np.mean(df[df['animal']==0.0], axis=0))\n",
    "\n",
    "# loop over all animals\n",
    "animal_avg = []\n",
    "for animal in animals:\n",
    "    animal_avg.append(df[df['animal']==animal].mean().values)\n",
    "animal_avg\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This series of steps is so common, that it has been implemented in pandas: [https://pandas.pydata.org/docs/user_guide/groupby.html]()\n",
    "\n",
    "It is applied in two steps: _groupby_ and _aggregate_\n",
    "1. _groupby_ groups the rows that have the same value in a specific column together\n",
    "2. _aggregate_ applies a computation that aggregates all data in a group to a single number, like the mean, standard deviation, or max.\n",
    "\n",
    "Here is an example of this in action. `df.groupby(column_name)` will produce a new object, that groups all rows with the same value in the specified column into a \"virtual subtable\":"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.groupby.generic.DataFrameGroupBy'>\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>animal</th>\n",
       "      <th>pre</th>\n",
       "      <th>post</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.211986</td>\n",
       "      <td>3.107796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.204871</td>\n",
       "      <td>2.830730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.560633</td>\n",
       "      <td>3.483980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.476412</td>\n",
       "      <td>2.801709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>3.662519</td>\n",
       "      <td>3.447005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.538696</td>\n",
       "      <td>3.301515</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   animal       pre      post\n",
       "0     0.0  2.211986  3.107796\n",
       "1     0.0  2.204871  2.830730\n",
       "2     0.0  2.560633  3.483980\n",
       "3     0.0  1.476412  2.801709\n",
       "4     0.0  3.662519  3.447005\n",
       "5     0.0  2.538696  3.301515"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped = df.groupby('animal')\n",
    "print(type(grouped))\n",
    "grouped.get_group(0)  # get the subtable for all rows where animal==0.0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Apply the \"mean\" operation to all values in each subtable and make a new table with the aggregate per-group data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/zr/6ql4dzjx0tq8mpzht_2dwh480000gn/T/ipykernel_3584/1479352327.py:1: FutureWarning: The provided callable <function mean at 0x1301a4a40> is currently using DataFrameGroupBy.mean. In a future version of pandas, the provided callable will be used directly. To keep current behavior pass the string \"mean\" instead.\n",
      "  agg = grouped.aggregate(np.mean)\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>pre</th>\n",
       "      <th>post</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>animal</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0.0</th>\n",
       "      <td>2.442519</td>\n",
       "      <td>3.162122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.0</th>\n",
       "      <td>2.738120</td>\n",
       "      <td>3.015360</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.0</th>\n",
       "      <td>3.363727</td>\n",
       "      <td>3.623694</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.0</th>\n",
       "      <td>2.321019</td>\n",
       "      <td>2.820372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4.0</th>\n",
       "      <td>2.037325</td>\n",
       "      <td>2.144355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5.0</th>\n",
       "      <td>1.981841</td>\n",
       "      <td>3.087824</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6.0</th>\n",
       "      <td>1.351306</td>\n",
       "      <td>1.552732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7.0</th>\n",
       "      <td>3.089612</td>\n",
       "      <td>3.516819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8.0</th>\n",
       "      <td>2.098225</td>\n",
       "      <td>2.791355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9.0</th>\n",
       "      <td>1.937971</td>\n",
       "      <td>2.033332</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             pre      post\n",
       "animal                    \n",
       "0.0     2.442519  3.162122\n",
       "1.0     2.738120  3.015360\n",
       "2.0     3.363727  3.623694\n",
       "3.0     2.321019  2.820372\n",
       "4.0     2.037325  2.144355\n",
       "5.0     1.981841  3.087824\n",
       "6.0     1.351306  1.552732\n",
       "7.0     3.089612  3.516819\n",
       "8.0     2.098225  2.791355\n",
       "9.0     1.937971  2.033332"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "agg = grouped.aggregate(np.mean)\n",
    "agg"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot the result:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 300x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "x = np.zeros_like(agg.values).T + [[0], [1]]\n",
    "y = agg.values.T\n",
    "\n",
    "plt.figure(figsize=(3, 6))\n",
    "plt.plot(x, y, 'o-k')\n",
    "plt.xticks([0, 1], labels=agg.columns)\n",
    "plt.ylabel('Score')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Example: Connectomics\n",
    "The connectome of the (female) fly brain was recently published.\n",
    "\n",
    "It can be accessed usign a beautiful web interface and user friendly web site: [https://codex.flywire.ai]().\n",
    "\n",
    "The web site is great for exploring the fly brain, but clicking around in a web browser has limitations in terms of the reproducibility and scale of an analysis.\n",
    "\n",
    "Therefore, the data was made availabel as a set of tables that can be loaded and processed with pandas.\n",
    "\n",
    "In the exercise, we will use pandas to answer neuroscience questions about the fly brain - here is a brief introduction to the data:\n",
    "\n",
    "There are three main tables:\n",
    "- connections - connections between cells\n",
    "- classification - cell classes (afferent, intrinsic, efferent), ...\n",
    "- (labels - human annotations of cell types)\n",
    "\n",
    "Each neuron has a unique identifier, the root_id. The root_id can be used to track neurons across the different tables.\n",
    "For instance, to find all neurons of a specific types, and then find their inputs or outputs."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classifications and connections\n",
    "The classification table has basic information about each of the 130k neurons in the fly brain: the cell_type, side of the brain, position in the brain etc."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>root_id</th>\n",
       "      <th>flow</th>\n",
       "      <th>super_class</th>\n",
       "      <th>class</th>\n",
       "      <th>sub_class</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>hemibrain_type</th>\n",
       "      <th>hemilineage</th>\n",
       "      <th>side</th>\n",
       "      <th>nerve</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>720575940627005443</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>optic</td>\n",
       "      <td>L1-5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>L4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>720575940615995398</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>optic</td>\n",
       "      <td>L1-5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>L4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>720575940621762567</td>\n",
       "      <td>afferent</td>\n",
       "      <td>ascending</td>\n",
       "      <td>AN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>CV</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>720575940624384007</td>\n",
       "      <td>afferent</td>\n",
       "      <td>sensory</td>\n",
       "      <td>olfactory</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>ORN_VA1v</td>\n",
       "      <td>NaN</td>\n",
       "      <td>left</td>\n",
       "      <td>AN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>720575940614422540</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>central</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>AOTU032,AOTU034</td>\n",
       "      <td>LALa1_posterior</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</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>127974</th>\n",
       "      <td>720575940621500404</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LLPC1,LLPC2a,LLPC2b,LLPC2c,LLPC2d,LLPC3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127975</th>\n",
       "      <td>720575940651646966</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>central</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LB0_anterior</td>\n",
       "      <td>center</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127976</th>\n",
       "      <td>720575940630151162</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>optic</td>\n",
       "      <td>optic_lobes</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127977</th>\n",
       "      <td>720575940629364731</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>optic</td>\n",
       "      <td>optic_lobes</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127978</th>\n",
       "      <td>720575940623859711</td>\n",
       "      <td>afferent</td>\n",
       "      <td>sensory</td>\n",
       "      <td>visual</td>\n",
       "      <td>NaN</td>\n",
       "      <td>R1-6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>127979 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   root_id       flow        super_class        class  \\\n",
       "0       720575940627005443  intrinsic              optic         L1-5   \n",
       "1       720575940615995398  intrinsic              optic         L1-5   \n",
       "2       720575940621762567   afferent          ascending           AN   \n",
       "3       720575940624384007   afferent            sensory    olfactory   \n",
       "4       720575940614422540  intrinsic            central          NaN   \n",
       "...                    ...        ...                ...          ...   \n",
       "127974  720575940621500404  intrinsic  visual_projection          NaN   \n",
       "127975  720575940651646966  intrinsic            central          NaN   \n",
       "127976  720575940630151162  intrinsic              optic  optic_lobes   \n",
       "127977  720575940629364731  intrinsic              optic  optic_lobes   \n",
       "127978  720575940623859711   afferent            sensory       visual   \n",
       "\n",
       "       sub_class cell_type                           hemibrain_type  \\\n",
       "0            NaN        L4                                      NaN   \n",
       "1            NaN        L4                                      NaN   \n",
       "2            NaN       NaN                                      NaN   \n",
       "3            NaN       NaN                                 ORN_VA1v   \n",
       "4            NaN       NaN                          AOTU032,AOTU034   \n",
       "...          ...       ...                                      ...   \n",
       "127974       NaN       NaN  LLPC1,LLPC2a,LLPC2b,LLPC2c,LLPC2d,LLPC3   \n",
       "127975       NaN       NaN                                      NaN   \n",
       "127976       NaN       NaN                                      NaN   \n",
       "127977       NaN       NaN                                      NaN   \n",
       "127978       NaN      R1-6                                      NaN   \n",
       "\n",
       "            hemilineage    side nerve  \n",
       "0                   NaN   right   NaN  \n",
       "1                   NaN   right   NaN  \n",
       "2                   NaN   right    CV  \n",
       "3                   NaN    left    AN  \n",
       "4       LALa1_posterior   right   NaN  \n",
       "...                 ...     ...   ...  \n",
       "127974              NaN   right   NaN  \n",
       "127975     LB0_anterior  center   NaN  \n",
       "127976              NaN    left   NaN  \n",
       "127977              NaN   right   NaN  \n",
       "127978              NaN   right   NaN  \n",
       "\n",
       "[127979 rows x 10 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classification = pd.read_csv('dat/flywire/classification.csv.gz')\n",
    "classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The connections table contains information about all ~4M synaptic synaptic connections:\n",
    "- pre_root_id: root_id of the presynaptic neuron, the source\n",
    "- post_root_id: root_id of the postsynaptic neuron, the target\n",
    "- syn_count: number of synapses in connection\n",
    "- nt_type: predicted neurotransmitter neurotransmitter \n",
    "- neuropil: target neuropil"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>pre_root_id</th>\n",
       "      <th>post_root_id</th>\n",
       "      <th>neuropil</th>\n",
       "      <th>syn_count</th>\n",
       "      <th>nt_type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>720575940596125868</td>\n",
       "      <td>720575940608552405</td>\n",
       "      <td>LOP_R</td>\n",
       "      <td>5</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>720575940596125868</td>\n",
       "      <td>720575940611348834</td>\n",
       "      <td>LOP_R</td>\n",
       "      <td>7</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>720575940596125868</td>\n",
       "      <td>720575940613059993</td>\n",
       "      <td>LOP_R</td>\n",
       "      <td>5</td>\n",
       "      <td>GLUT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>720575940596125868</td>\n",
       "      <td>720575940616986553</td>\n",
       "      <td>LOP_R</td>\n",
       "      <td>5</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>720575940596125868</td>\n",
       "      <td>720575940620124326</td>\n",
       "      <td>LOP_R</td>\n",
       "      <td>8</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3794610</th>\n",
       "      <td>720575940660868737</td>\n",
       "      <td>720575940607206786</td>\n",
       "      <td>ME_L</td>\n",
       "      <td>9</td>\n",
       "      <td>GABA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3794611</th>\n",
       "      <td>720575940660868737</td>\n",
       "      <td>720575940608664873</td>\n",
       "      <td>ME_L</td>\n",
       "      <td>6</td>\n",
       "      <td>GABA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3794612</th>\n",
       "      <td>720575940660868737</td>\n",
       "      <td>720575940611462242</td>\n",
       "      <td>ME_L</td>\n",
       "      <td>6</td>\n",
       "      <td>GABA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3794613</th>\n",
       "      <td>720575940660868737</td>\n",
       "      <td>720575940622913063</td>\n",
       "      <td>ME_L</td>\n",
       "      <td>23</td>\n",
       "      <td>GABA</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3794614</th>\n",
       "      <td>720575940660868737</td>\n",
       "      <td>720575940626553546</td>\n",
       "      <td>ME_L</td>\n",
       "      <td>6</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3794615 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                pre_root_id        post_root_id neuropil  syn_count nt_type\n",
       "0        720575940596125868  720575940608552405    LOP_R          5     ACH\n",
       "1        720575940596125868  720575940611348834    LOP_R          7     ACH\n",
       "2        720575940596125868  720575940613059993    LOP_R          5    GLUT\n",
       "3        720575940596125868  720575940616986553    LOP_R          5     ACH\n",
       "4        720575940596125868  720575940620124326    LOP_R          8     ACH\n",
       "...                     ...                 ...      ...        ...     ...\n",
       "3794610  720575940660868737  720575940607206786     ME_L          9    GABA\n",
       "3794611  720575940660868737  720575940608664873     ME_L          6    GABA\n",
       "3794612  720575940660868737  720575940611462242     ME_L          6    GABA\n",
       "3794613  720575940660868737  720575940622913063     ME_L         23    GABA\n",
       "3794614  720575940660868737  720575940626553546     ME_L          6     ACH\n",
       "\n",
       "[3794615 rows x 5 columns]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "connections = pd.read_csv('dat/flywire/connections.csv.gz')\n",
    "connections"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Example: Find neurons of a specific cell type\n",
    "LC10a is a cluster of higher-order visual neurons required by Drosophila males to track the female during courtship.\n",
    "\n",
    "We can find all LC10a neurons in the fly brain using boolean indexing into the classification table:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>root_id</th>\n",
       "      <th>flow</th>\n",
       "      <th>super_class</th>\n",
       "      <th>class</th>\n",
       "      <th>sub_class</th>\n",
       "      <th>cell_type</th>\n",
       "      <th>hemibrain_type</th>\n",
       "      <th>hemilineage</th>\n",
       "      <th>side</th>\n",
       "      <th>nerve</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3035</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3691</th>\n",
       "      <td>720575940638285109</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3737</th>\n",
       "      <td>720575940614430111</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4456</th>\n",
       "      <td>720575940623082237</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>right</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4544</th>\n",
       "      <td>720575940623868853</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</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>125773</th>\n",
       "      <td>720575940631195217</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126409</th>\n",
       "      <td>720575940631982929</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126528</th>\n",
       "      <td>720575940631983185</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126843</th>\n",
       "      <td>720575940627265223</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127467</th>\n",
       "      <td>720575940610227162</td>\n",
       "      <td>intrinsic</td>\n",
       "      <td>visual_projection</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>LC10a</td>\n",
       "      <td>LC10</td>\n",
       "      <td>VPNd2</td>\n",
       "      <td>left</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>227 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                   root_id       flow        super_class class sub_class  \\\n",
       "3035    720575940620720112  intrinsic  visual_projection   NaN       NaN   \n",
       "3691    720575940638285109  intrinsic  visual_projection   NaN       NaN   \n",
       "3737    720575940614430111  intrinsic  visual_projection   NaN       NaN   \n",
       "4456    720575940623082237  intrinsic  visual_projection   NaN       NaN   \n",
       "4544    720575940623868853  intrinsic  visual_projection   NaN       NaN   \n",
       "...                    ...        ...                ...   ...       ...   \n",
       "125773  720575940631195217  intrinsic  visual_projection   NaN       NaN   \n",
       "126409  720575940631982929  intrinsic  visual_projection   NaN       NaN   \n",
       "126528  720575940631983185  intrinsic  visual_projection   NaN       NaN   \n",
       "126843  720575940627265223  intrinsic  visual_projection   NaN       NaN   \n",
       "127467  720575940610227162  intrinsic  visual_projection   NaN       NaN   \n",
       "\n",
       "       cell_type hemibrain_type hemilineage   side nerve  \n",
       "3035       LC10a           LC10       VPNd2  right   NaN  \n",
       "3691       LC10a           LC10       VPNd2   left   NaN  \n",
       "3737       LC10a           LC10       VPNd2  right   NaN  \n",
       "4456       LC10a           LC10       VPNd2  right   NaN  \n",
       "4544       LC10a           LC10       VPNd2   left   NaN  \n",
       "...          ...            ...         ...    ...   ...  \n",
       "125773     LC10a           LC10       VPNd2   left   NaN  \n",
       "126409     LC10a           LC10       VPNd2   left   NaN  \n",
       "126528     LC10a           LC10       VPNd2   left   NaN  \n",
       "126843     LC10a           LC10       VPNd2   left   NaN  \n",
       "127467     LC10a           LC10       VPNd2   left   NaN  \n",
       "\n",
       "[227 rows x 10 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "LC10a_neurons = classification[classification['cell_type']=='LC10a']\n",
    "LC10a_neurons"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Example: Find all postsynaptic partners of a given neuron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "720575940620720112\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>pre_root_id</th>\n",
       "      <th>post_root_id</th>\n",
       "      <th>neuropil</th>\n",
       "      <th>syn_count</th>\n",
       "      <th>nt_type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1309977</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940610389873</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>10</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309978</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940613946783</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>10</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309979</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940614041238</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>9</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309980</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940616012061</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>26</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309981</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940616169629</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>14</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309982</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940616888020</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>7</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309983</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940619453861</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>5</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309984</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940620321158</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>11</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309985</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940621925631</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>7</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309986</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940622538520</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>24</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309987</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940623107509</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>5</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309988</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940626024336</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>5</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309989</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940626643658</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>6</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309990</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940627738640</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>22</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309991</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940628916444</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>13</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309992</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940628972538</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>8</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309993</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940629883971</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>17</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309994</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940631517251</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>31</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309995</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940638668659</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>11</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1309996</th>\n",
       "      <td>720575940620720112</td>\n",
       "      <td>720575940652922529</td>\n",
       "      <td>AOTU_R</td>\n",
       "      <td>11</td>\n",
       "      <td>ACH</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                pre_root_id        post_root_id neuropil  syn_count nt_type\n",
       "1309977  720575940620720112  720575940610389873   AOTU_R         10     ACH\n",
       "1309978  720575940620720112  720575940613946783   AOTU_R         10     ACH\n",
       "1309979  720575940620720112  720575940614041238   AOTU_R          9     ACH\n",
       "1309980  720575940620720112  720575940616012061   AOTU_R         26     ACH\n",
       "1309981  720575940620720112  720575940616169629   AOTU_R         14     ACH\n",
       "1309982  720575940620720112  720575940616888020   AOTU_R          7     ACH\n",
       "1309983  720575940620720112  720575940619453861   AOTU_R          5     ACH\n",
       "1309984  720575940620720112  720575940620321158   AOTU_R         11     ACH\n",
       "1309985  720575940620720112  720575940621925631   AOTU_R          7     ACH\n",
       "1309986  720575940620720112  720575940622538520   AOTU_R         24     ACH\n",
       "1309987  720575940620720112  720575940623107509   AOTU_R          5     ACH\n",
       "1309988  720575940620720112  720575940626024336   AOTU_R          5     ACH\n",
       "1309989  720575940620720112  720575940626643658   AOTU_R          6     ACH\n",
       "1309990  720575940620720112  720575940627738640   AOTU_R         22     ACH\n",
       "1309991  720575940620720112  720575940628916444   AOTU_R         13     ACH\n",
       "1309992  720575940620720112  720575940628972538   AOTU_R          8     ACH\n",
       "1309993  720575940620720112  720575940629883971   AOTU_R         17     ACH\n",
       "1309994  720575940620720112  720575940631517251   AOTU_R         31     ACH\n",
       "1309995  720575940620720112  720575940638668659   AOTU_R         11     ACH\n",
       "1309996  720575940620720112  720575940652922529   AOTU_R         11     ACH"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# select the first LC10a neuron in the list and get it's root_id\n",
    "pre_root_id = LC10a_neurons.iloc[0]['root_id']\n",
    "print(pre_root_id)\n",
    "\n",
    "# find all downstream targets for that neuron\n",
    "# that is, all synaptic connections for which our neuron's root_id is the root_id of the pre-synapse\n",
    "targets = connections[connections['pre_root_id']==pre_root_id]\n",
    "targets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernel_info": {
   "name": "python3"
  },
  "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": "3.12.0"
  },
  "nteract": {
   "version": "0.12.3"
  },
  "vscode": {
   "interpreter": {
    "hash": "7ea0ec616133ead53c1908c8f6539f5c0cb9b2f78368e2bb6ab3f847e89ca400"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}