Annotations of Pandas DataFrame

escola.xlsx

FromDict.ipynb

gen_names_xlsx.ipynb

gen_random_values.ipynb

import_xlsx_to_django_database.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import timeit\n",
    "import xlrd\n",
    "from django.contrib.auth.models import User"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "tic = timeit.default_timer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "filename = '/tmp/example.xlsx'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_excel(filename)"
   ]
  },
  {
   "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>NAME</th>\n",
       "      <th>EMAIL</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2011</th>\n",
       "      <td>Angela Brown</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2824</th>\n",
       "      <td>Angela Brown</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3599</th>\n",
       "      <td>Angela Jones</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1227</th>\n",
       "      <td>Angela Jones</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2605</th>\n",
       "      <td>Anthony Evans</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               NAME                    EMAIL\n",
       "2011   Angela Brown   [email protected]\n",
       "2824   Angela Brown   [email protected]\n",
       "3599   Angela Jones   [email protected]\n",
       "1227   Angela Jones   [email protected]\n",
       "2605  Anthony Evans  [email protected]"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "email = df['EMAIL']\n",
    "# Retornando os itens duplicados\n",
    "df[email.isin(email[email.duplicated()])].sort_values(by=['EMAIL']).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>index</th>\n",
       "      <th>NAME</th>\n",
       "      <th>EMAIL</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4957</th>\n",
       "      <td>4995</td>\n",
       "      <td>Helen Mcallister</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4958</th>\n",
       "      <td>4996</td>\n",
       "      <td>Scott Hall</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4959</th>\n",
       "      <td>4997</td>\n",
       "      <td>Dawn Dowling</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4960</th>\n",
       "      <td>4998</td>\n",
       "      <td>John Campbell</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4961</th>\n",
       "      <td>4999</td>\n",
       "      <td>Barbara Alldredge</td>\n",
       "      <td>[email protected]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      index               NAME                        EMAIL\n",
       "4957   4995   Helen Mcallister   [email protected]\n",
       "4958   4996         Scott Hall         [email protected]\n",
       "4959   4997       Dawn Dowling       [email protected]\n",
       "4960   4998      John Campbell      [email protected]\n",
       "4961   4999  Barbara Alldredge  [email protected]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Removendo os itens duplicados\n",
    "dffinal = df.drop_duplicates('EMAIL').reset_index()\n",
    "dffinal.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_user(new_users):\n",
    "    # Separa full_name em first_name e last_name\n",
    "    users = []\n",
    "    for new_user in new_users:\n",
    "        full_name = new_user['full_name']\n",
    "        email = new_user['email']\n",
    "        first_name = full_name.split()[0]\n",
    "        last_name = full_name.split()[1:]\n",
    "        last_name = ' '.join(last_name)\n",
    "        username = email\n",
    "        user = User(\n",
    "            first_name=first_name,\n",
    "            last_name=last_name,\n",
    "            email=email,\n",
    "            username=username\n",
    "        )\n",
    "        users.append(user)\n",
    "    return users"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def import_users(new_users):\n",
    "    User.objects.bulk_create(new_users)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_users = []\n",
    "emails = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "emails = dffinal['EMAIL'].values.tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "exist_users = User.objects.filter(email__in=emails).values_list('email', flat=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Manuel Marshall [email protected]\n",
      "Joy Flores [email protected]\n",
      "Nicholas Waits [email protected]\n",
      "Laura Cross [email protected]\n",
      "Felipe Garland [email protected]\n"
     ]
    }
   ],
   "source": [
    "for row in dffinal.head().itertuples():\n",
    "    print(row.NAME, row.EMAIL)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "for row in dffinal.itertuples():\n",
    "    full_name = row.NAME\n",
    "    email = row.EMAIL\n",
    "    if email not in exist_users:\n",
    "        data = dict(full_name=full_name, email=email)\n",
    "        new_users.append(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_users_ = create_user(new_users)\n",
    "import_users(new_users_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "17 existentes\n",
      "4945 novos\n"
     ]
    }
   ],
   "source": [
    "print(len(exist_users), 'existentes')\n",
    "print(len(new_users), 'novos')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "toc = timeit.default_timer()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.1476561470044544"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "toc - tic"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Django Shell-Plus",
   "language": "python",
   "name": "django_extensions"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}

import_xlsx_to_django_database2.ipynb

OrcamentosNumOrcNovo.ipynb

pandas.ipynb

remove_duplicated.ipynb

separe_email.ipynb

Código que substitui da célula [6] em diante do separe_email.ipynb:

emails_df = df.email_.str.split(";").apply(pd.Series).stack().reset_index(level=1, drop=True).to_frame('emails')

dffinal = df.drop("email_", axis=1).merge(emails_df, left_index=True, right_index=True)

Um detalhe é a falta de um índice único dos registros. Então, supondo o índice automático decorrente da importação do arquivo Excel como ID válido, esse ID é o usado no "join" (merge).

Para transformar Query do Django em DataFrame:
df = pd.DataFrame(list(User.objects.all().values()))

Pra mostrar somente algumas colunas:
df[['id', 'email']]

Transformando todos os campos do DataFrame num jSON
df.T.apply(dict).tolist()

Renomeando colunas
df.rename(columns={'old_name': 'new_name'})

Selecionando as colunas de 10 em 10
df.iloc[:, :10]
com transposição
df.T.iloc[:10]

Novo df com algumas colunas.
new = old.filter(['A','B','D'], axis=1)

Como não inserir chaves com valor nulo no dicionário:
df.T.apply(lambda x: dict(x.dropna())).tolist()

Retornando os valores cujo tamanho da string seja maior que...
df[df['foo'].str.len() > 50]['foo']

Mostrando o tamanho de cada célula
df[df['foo'].str.len() > 50]['foo'].str.len()

Mostrando o tamanho de cada célula da coluna

df['foo_len'] = df['foo'].apply(len)
df[['foo_len', 'foo']]

Ou

df['foo'].str.len()

Suponha que você tenha Cliente e Obra.

Pegando o ID do Cliente que está em Obra e trocando pelo nome do Cliente que está no outro DataFrame.

# JSON do Cliente com IDCliente e Cliente
dict_cliente = df_cliente[['IDCliente', 'Cliente']].T.apply(dict).tolist()

[
 {'IDCliente': 288, 'Cliente': 'Cliente Um'},
 {'IDCliente': 1, 'Cliente': 'Cliente Dois'},
 {'IDCliente': 959, 'Cliente': 'Cliente Três'},
]

Montando o dicionário que será usado como busca de cada Cliente a partir do seu ID.

_dict_cliente = {}
for item in dict_cliente:
    _dict_cliente[item['IDCliente']] = item['Cliente']

_dict_cliente

{
 288: 'Cliente Um',
 1: 'Cliente Dois',
 959: 'Cliente Três',
}

A partir desse dicionário fazemos a busca no outro DataFrame.

for row in df.itertuples():
    nome_cliente = _dict_cliente.get(row.IDCliente)
    print(row.IDCliente, nome_cliente)

288 Cliente Um
1   Cliente Dois
959 Cliente Três

Retornando o valor máximo agrupado por ano.

df.groupby(['Ano'])['NumeroOrcamento'].max()
df.reset_index()

Verificando data vazia:

for row in df.itertuples():
    if row.DataOrcamento is pd.NaT:
        print('Vazio')
    else:
        print(row.DataOrcamento)

Definindo vários fillna diferentes por coluna:

values = {'last_name': '', 'occupation': '', 'age': 0}
df = df.fillna(value=values)
df.head()

Se tiver problema com liblzma, faça um downgrade do Pandas para pandas==0.24.2.
https://stackoverflow.com/a/57115325

Retorna o tamanho do maior objeto de cada coluna.

dict_sizes = {}
for col in df.columns:
    try:
        print(f'{col} max length: {df[col].map(len).max()}\n')
        dict_sizes[col] = df[col].map(len).max()
    except Exception as e:
        raise e
dict_sizes

from_dict doc

dtype example
df['estoque'] =df['estoque'].fillna(0).astype(int)

Pandas Dataframe df to Django
https://www.laurivan.com/save-pandas-dataframe-as-django-model/

Produto.objects.bulk_create(
    Produto(**item) for item in df.to_dict('records')
)

Definindo os tipos das colunas com dtype

dict_types_annot = {
    'produto': str,
    'ncm': str,
    'preco': float,
    'estoque': 'Int64',
}

# Define os tipos das colunas
dff = df.astype(dict_types_annot, errors='ignore')

# Troca 'nan' por None e float por None.
dff = dff.replace({'nan': None, float('nan'): None})

dff.to_dict('records')

Produto.objects.bulk_create(
    Produto(**item) for item in dff.to_dict('records')
)

Intersecção de dataframes

import pandas as pd
import numpy as np
import datetime
from random import randint

df1 = pd.DataFrame({
    'letters': ['A', 'B', 'C', 'D', 'E', 'J', 'K', 'M'],
    'B': np.random.randint(0, 10, 8),
})
df1

df2 = pd.DataFrame({
    'letters': ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'U', 'Z'],
    'B': np.random.randint(0, 10, 26),
})
df2

# Retorna o que tem de comum nos dois dataframes.
pd.merge(df1, df2, how='inner', on='fruits')

# Retorna o que tem de comum, considerando o df1.
pd.merge(df1, df2, how='left', on='fruits')

# Retorna o que tem de comum, considerando o df2.
pd.merge(df1, df2, how='right', on='fruits')