pre-process_2011.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "22f4a66e-05dd-47f9-8cd3-d27349640c96",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "23852f68-ccb4-453a-a7a3-c91ec492f002",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11 = pd.read_csv('computacao_enade_2011.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b536e8c4-4aa9-4f38-81da-00a36840e84d",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11.replace({'CO_GRUPO': {72: 'ADS', 4004: 'BCC', 5809: 'EC', 4005: 'LCC', 79: 'RC', 4006: 'SI'}}, inplace=True) # inplace=True permite que o df seja modificado diretamente\n",
    "\n",
    "# dft1 = dft1.astype({\"a\": np.bool_, \"c\": np.float64})\n",
    "dados_11 = dados_11.astype({'CO_GRUPO':str})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "3b28d855-2d35-4b48-b2b3-a1ef2e21ae0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "estado_dict = {\n",
    "11 : 'RO',12 : 'AC', \n",
    "13 : 'AM',14 : 'RR',\n",
    "15 : 'PA',16 : 'AP',\n",
    "17 : 'TO',21 : 'MA',\n",
    "22 : 'PI',23 : 'CE',\n",
    "24 : 'RN',25 : 'PB',\n",
    "26 : 'PE',27 : 'AL',\n",
    "28 : 'SE',29 : 'BA',\n",
    "31 : 'MG',32 : 'ES',\n",
    "33 : 'RJ',35 : 'SP',\n",
    "41 : 'PR',42 : 'SC',\n",
    "43 : 'RS',50 : 'MS',\n",
    "51 : 'MT',52 : 'GO', \n",
    "53 : 'DF'\n",
    "}\n",
    "\n",
    "dados_11['CO_UF_CURSO'].replace(estado_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'CO_UF_CURSO':str})\n",
    "\n",
    "regiao_dict = {\n",
    "1 : 'N',2 : 'NE', \n",
    "3 : 'SE',4 : 'SUL',\n",
    "5 : 'CO'\n",
    "}\n",
    "\n",
    "dados_11['CO_REGIAO_CURSO'].replace(regiao_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'CO_REGIAO_CURSO':str})\n",
    "\n",
    "# Não há IESes do tipo especial\n",
    "\n",
    "# ies_especial = dados_11[dados_11[\"CO_CATEGAD\"] == 7]\n",
    "# ies_especial\n",
    "\n",
    "cat_adm_dict = {\n",
    "1 : 'Pública Federal',2 : 'Pública Estadual', \n",
    "3 : 'Pública Municipal',4 : 'Privada CFL',\n",
    "5 : 'Privada SFL'\n",
    "}\n",
    "\n",
    "dados_11['CO_CATEGAD'].replace(cat_adm_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'CO_CATEGAD':str})\n",
    "\n",
    "# cefet = dados_11[dados_11['CO_ORGACAD'] == 10019]\n",
    "# cefet\n",
    "\n",
    "cat_org_dict = {\n",
    "10019 : 'CEFET',10020 : 'Centro Universitário', \n",
    "10022 : 'Faculdade',10026 : 'IFECT',\n",
    "10028 : 'Universidade'\n",
    "}\n",
    "\n",
    "dados_11['CO_ORGACAD'].replace(cat_org_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'CO_ORGACAD':str})\n",
    "\n",
    "dados_11[\"CO_TURNO_GRADUACAO\"] = dados_11['IN_MATUT'] * 1 + dados_11['IN_VESPER'] * 3 + dados_11['IN_NOTURNO'] * 5\n",
    "\n",
    "turno_dict = {\n",
    "1 : 'Matutino', 3 : 'Vespertino', \n",
    "5 : 'Noturno', 4 : 'Integral',\n",
    "8 : 'Integral', 6 : 'Integral',\n",
    "9 : 'Integral'\n",
    "}\n",
    "\n",
    "dados_11['CO_TURNO_GRADUACAO'].replace(turno_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'CO_TURNO_GRADUACAO':str})\n",
    "\n",
    "dados_11.drop(columns=['IN_MATUT', 'IN_VESPER', 'IN_NOTURNO'], inplace=True)\n",
    "\n",
    "col_turno = dados_11.pop('CO_TURNO_GRADUACAO')       # Muda \"CO_TURNO_GRADUACAO\" para sétima posição\n",
    "dados_11.insert(6, col_turno.name, col_turno)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "101e5c7f-a13b-4d16-87bb-7f736c3d096d",
   "metadata": {},
   "source": [
    "### Análise dos atributos ANO_FIM_EM e ANO_IN_GRAD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bbf6a1b5-ca0c-4ffa-8c59-638946e74d54",
   "metadata": {},
   "outputs": [],
   "source": [
    "outliers_afem = dados_11[(dados_11['ANO_FIM_EM'] < 1951) | (dados_11['ANO_FIM_EM'] > 2011)]\n",
    "outliers_afem\n",
    "\n",
    "outliers_aig = dados_11[(dados_11['ANO_IN_GRAD'] < 1971) | (dados_11['ANO_IN_GRAD'] > 2011)]\n",
    "outliers_aig"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d34947d9-cbe6-4249-a124-1d5ab33ce671",
   "metadata": {},
   "outputs": [],
   "source": [
    "media_afem = dados_11['ANO_FIM_EM'].mean()\n",
    "media_afem\n",
    "\n",
    "media_aig = dados_11['ANO_IN_GRAD'].mean()\n",
    "media_aig"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "be9ee252-4eec-4419-9f98-babee1a5ee07",
   "metadata": {},
   "source": [
    "### ---------------------------------------------------------------------------------------------"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "32f253a5-d9bb-4890-8eaa-48dd4c40668a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Atributo derivado 1 - DIF_GRAD_EM\n",
    "dados_11[\"DIF_GRAD_EM\"] = dados_11['ANO_IN_GRAD'] - dados_11['ANO_FIM_EM']\n",
    "\n",
    "# Atributo derivado 2 - DIF_ANO_GRAD\n",
    "dados_11[\"DIF_ANO_GRAD\"] = (dados_11['NU_ANO'] - dados_11['ANO_IN_GRAD']) + 1\n",
    "\n",
    "# Ajuste dos valores em \"ANO_FIM_EM\"\n",
    "dados_11[\"ANO_FIM_EM\"] = dados_11['NU_ANO'] - dados_11['ANO_FIM_EM']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "4a5d41f9-4be5-42aa-8c2b-0c999d94241d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "747"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Ver o que acontece em dados_11[dados_11[\"QE_I17\"] == \"E\"]\n",
    "\n",
    "dados_11_Privada_Publica = dados_11[dados_11[\"QE_I17\"] == \"E\"]\n",
    "len_Tipo_EM_E = len(dados_11_Privada_Publica)\n",
    "\n",
    "len_Tipo_EM_E"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a74bf180-18c9-4888-8f9e-ed3628b57d45",
   "metadata": {},
   "outputs": [],
   "source": [
    "metade_len = len_Tipo_EM_E / 2\n",
    "\n",
    "count = 0\n",
    "turn = 0\n",
    "\n",
    "# Metade das linhas que contém em \"QE_I17\" o valor \"E\" passam a valer \"Pública\", e a outra metade \"Privada\"\n",
    "for i, j in dados_11.iterrows():\n",
    "    if count >= metade_len:\n",
    "        turn = 1;\n",
    "  \n",
    "    if dados_11.at[i,\"QE_I17\"] == \"E\":\n",
    "        if not(turn):\n",
    "            dados_11.at[i,\"QE_I17\"] = \"Pública\"\n",
    "        else:\n",
    "            dados_11.at[i,\"QE_I17\"] = \"Privada\"\n",
    "            \n",
    "        count += 1\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "cbc0c413-59fa-448d-b1fc-b40581d9a37c",
   "metadata": {},
   "outputs": [],
   "source": [
    "tipo_em_dict = {\n",
    "'A' : 'Pública','B' : 'Privada', \n",
    "'C' : 'Exterior', 'D' : 'Pública',\n",
    "'E' :'Privada', 'F': 'Exterior'\n",
    "}\n",
    "\n",
    "dados_11['QE_I17'].replace(tipo_em_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I17':str})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "f92916e0-1653-4f8d-8724-5bf51d818d02",
   "metadata": {},
   "outputs": [],
   "source": [
    "mod_em_dict = {\n",
    "'A' : 'Tradicional','B' : 'Técnico', \n",
    "'C' : 'Magistério', 'D' : 'Supletivo',\n",
    "'E':'Outra'\n",
    "}\n",
    "\n",
    "dados_11['QE_I18'].replace(mod_em_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I18':str})\n",
    "\n",
    "raca_dict = {\n",
    "'A' : 'Branca','B' : 'Preta', \n",
    "'C' : 'Parda', 'D' : 'Amarela',\n",
    "'E': 'Indígena', 'nan': 'Não declarado'\n",
    "}\n",
    "\n",
    "dados_11['QE_I02'].replace(raca_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I02':str})\n",
    "\n",
    "est_civil_dict = {\n",
    "'A' : 'Solteiro','B' : 'Casado', \n",
    "'C' : 'Separado', 'D' : 'Viúvo',\n",
    "'E': 'Outro'\n",
    "}\n",
    "\n",
    "dados_11['QE_I01'].replace(est_civil_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I01':str})\n",
    "\n",
    "nl_pai_mae_dict = {\n",
    "'A' : 'Nenhuma','B' : 'Fundamental 1-5', \n",
    "'C' : 'Fundamental 6-9', 'D' : 'Médio',\n",
    "'E': 'Superior', 'F': 'Pós-graduação'\n",
    "}\n",
    "\n",
    "dados_11['QE_I13'].replace(nl_pai_mae_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I13':str})\n",
    "\n",
    "dados_11['QE_I14'].replace(nl_pai_mae_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I14':str})\n",
    "\n",
    "uf_em_dict = {\n",
    "11 : 'RO',12 : 'AC', \n",
    "13 : 'AM',14 : 'RR',\n",
    "15 : 'PA',16 : 'AP',\n",
    "17 : 'TO',21 : 'MA',\n",
    "22 : 'PI',23 : 'CE',\n",
    "24 : 'RN',25 : 'PB',\n",
    "26 : 'PE',27 : 'AL',\n",
    "28 : 'SE',29 : 'BA',\n",
    "31 : 'MG',32 : 'ES',\n",
    "33 : 'RJ',35 : 'SP',\n",
    "41 : 'PR',42 : 'SC',\n",
    "43 : 'RS',50 : 'MS',\n",
    "51 : 'MT',52 : 'GO', \n",
    "53 : 'DF',99 : 'Não se aplica'\n",
    "}\n",
    "\n",
    "dados_11['QE_I15'].replace(uf_em_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I15':str})\n",
    "\n",
    "renda_dict = {\n",
    "'A' : 'Até 1,5 SM','B' : 'Até 1,5 SM', \n",
    "'C' : 'De 1,5 a 3 SM', 'D' : 'De 3 a 4,5 SM',\n",
    "'E': 'De 4,5 a 6 SM', 'F': 'De 6 a 10 SM',\n",
    "'G': 'De 10 a 30 SM', 'H': 'Acima de 30 SM'\n",
    "}\n",
    "\n",
    "dados_11['QE_I05'].replace(renda_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I05':str})\n",
    "\n",
    "status_financ_dict = {\n",
    "'A' : 'Sem renda/Outros', 'B' : 'Tem renda/Outros', \n",
    "'C' : 'Tem renda/Independente', 'D' : 'Tem renda/Sustenta', \n",
    "'E' : 'Tem renda/Principal'\n",
    "}\n",
    "\n",
    "dados_11['QE_I06'].replace(status_financ_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I06':str})\n",
    "\n",
    "moradia_dict = {\n",
    "'A' : 'Sozinho','B' : 'Parentes', \n",
    "'C' : 'Cônjuge/Filhos', 'D' : 'Outras pessoas',\n",
    "'E': 'Universidade', 'F': 'Habitação individual/coletiva'\n",
    "}\n",
    "\n",
    "dados_11['QE_I03'].replace(moradia_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I03':str})\n",
    "\n",
    "qtde_moradia_dict = {\n",
    "'A' : 'Nenhuma','B' : 'Uma', \n",
    "'C' : 'Duas', 'D' : 'Três',\n",
    "'E': 'Quatro', 'F': 'Cinco',\n",
    "'G': 'Seis', 'H': 'Sete ou mais'\n",
    "}\n",
    "\n",
    "dados_11['QE_I04'].replace(qtde_moradia_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I04':str})\n",
    "\n",
    "politica_dict = {\n",
    "'A' : 'Não','B' : 'Critério étnico-racial',\n",
    "'C' : 'Critério de renda', 'D' : 'Bolsa de estudos',\n",
    "'E': 'Combinação de critérios', 'F': 'Critérios diferentes'\n",
    "}\n",
    "\n",
    "dados_11['QE_I12'].replace(politica_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I12':str})\n",
    "\n",
    "trabalho_dict = {\n",
    "'A' : 'Não','B' : 'Eventualmente',\n",
    "'C' : 'Até 20 horas', 'D' : '21 a 39 horas',\n",
    "'E': '40 horas ou mais'\n",
    "}\n",
    "\n",
    "dados_11['QE_I07'].replace(trabalho_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I07':str})\n",
    "\n",
    "recebeu_bolsa_dict = {\n",
    "'A' : 'Sim','B' : 'Não se aplica',\n",
    "'C' : 'Não'\n",
    "}\n",
    "\n",
    "dados_11['QE_I09'].replace(recebeu_bolsa_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I09':str})\n",
    "\n",
    "qtde_livros_dict = {\n",
    "'A' : 'Nenhum','B' : '1 ou 2',\n",
    "'C' : '3-5', 'D' : '6-8',\n",
    "'E': '+8'\n",
    "}\n",
    "\n",
    "dados_11['QE_I19'].replace(qtde_livros_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I19':str})\n",
    "\n",
    "horas_estudo_dict = {\n",
    "'A' : 'Nenhum','B' : '1-3',\n",
    "'C' : '4-7', 'D' : '8-12',\n",
    "'E': '+12'\n",
    "}\n",
    "\n",
    "dados_11['QE_I20'].replace(horas_estudo_dict,inplace=True)\n",
    "dados_11 = dados_11.astype({'QE_I20':str})\n",
    "\n",
    "cond_salas_dict = {\n",
    "    'A' : 'Todas',\n",
    "    'B' : 'Maior parte',\n",
    "    'C' : 'Algumas',\n",
    "    'D' : 'Nenhuma'\n",
    "}\n",
    "\n",
    "dados_11['QE_I22'].replace(cond_salas_dict,inplace=True)\n",
    "dados_11['QE_I22'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I22':str})\n",
    "\n",
    "cond_praticas_dict = {\n",
    "    'A' : 'Todas',\n",
    "    'B' : 'Maior parte',\n",
    "    'C' : 'Algumas',\n",
    "    'D' : 'Nenhuma'\n",
    "}\n",
    "\n",
    "dados_11['QE_I26'].replace(cond_praticas_dict,inplace=True)\n",
    "dados_11['QE_I26'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I26':str})\n",
    "\n",
    "disp_professores_dict = {\n",
    "    'A' : 'Todos',\n",
    "    'B' : 'Maior parte',\n",
    "    'C' : 'Alguns',\n",
    "    'D' : 'Nenhum'\n",
    "}\n",
    "\n",
    "dados_11['QE_I41'].replace(disp_professores_dict,inplace=True)\n",
    "dados_11['QE_I41'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I41':str})\n",
    "\n",
    "dominio_professores_dict = {\n",
    "    'A' : 'Todos',\n",
    "    'B' : 'Maior parte',\n",
    "    'C' : 'Alguns',\n",
    "    'D' : 'Nenhum'\n",
    "}\n",
    "\n",
    "dados_11['QE_I42'].replace(dominio_professores_dict,inplace=True)\n",
    "dados_11['QE_I42'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I42':str})\n",
    "\n",
    "plano_ensino_dict = {\n",
    "    'A' : 'Amplamente',\n",
    "    'B' : 'Parcialmente',\n",
    "    'C' : 'Muito pouco',\n",
    "    'D' : 'Não contribui'\n",
    "}\n",
    "\n",
    "dados_11['QE_I52'].replace(plano_ensino_dict,inplace=True)\n",
    "dados_11['QE_I52'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I52':str})\n",
    "\n",
    "formacao_dict = {\n",
    "    'A' : 'Muito boa',\n",
    "    'B' : 'Boa',\n",
    "    'C' : 'Regular',\n",
    "    'D' : 'Fraca',\n",
    "    'E' : 'Muito fraca'\n",
    "}\n",
    "\n",
    "dados_11['QE_I54'].replace(formacao_dict,inplace=True)\n",
    "dados_11['QE_I54'].fillna(\"Não sei responder\", inplace = True)\n",
    "dados_11 = dados_11.astype({'QE_I54':str})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "900a6074-f2b5-4c14-b805-46f56e3375ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "columns_dict = {\n",
    "    \"NU_ANO\": \"Ano\",\n",
    "    \"CO_CURSO\": \"Codigo_do_Curso\",\n",
    "    \"CO_GRUPO\": \"Curso\",\n",
    "    \"CO_UF_CURSO\": \"UF\",\n",
    "    \"CO_REGIAO_CURSO\": \"Regiao\",\n",
    "    \"CO_CATEGAD\": \"Cat_Administrativa\",\n",
    "    \"CO_ORGACAD\": \"Cat_Org_Academica\",\n",
    "    \"CO_TURNO_GRADUACAO\": \"Turno\",\n",
    "    \"TP_PRES\": \"Tipo_Presenca\",\n",
    "    \"NT_FG\": \"Nota_FG\",\n",
    "    \"NT_CE\": \"Nota_CE\",\n",
    "    \"DIF_GRAD_EM\": \"Anos_Preparacao\",\n",
    "    \"DIF_ANO_GRAD\": \"Anos_Graduacao\",\n",
    "    \"ANO_FIM_EM\": \"Conclusao_Ensino_Medio\",\n",
    "    \"ANO_IN_GRAD\": \"Inicio_Graduacao\",\n",
    "    \"QE_I17\": \"Tipo_Escola_EM\",\n",
    "    \"QE_I18\": \"Modalidade_EM\",\n",
    "    \"NU_IDADE\": \"Idade\",\n",
    "    \"TP_SEXO\": \"Sexo\",\n",
    "    \"QE_I02\": \"Cor_Raca\",\n",
    "    \"QE_I01\": \"Estado_Civil\",\n",
    "    \"QE_I13\": \"Nl_Pai\",\n",
    "    \"QE_I14\": \"Nl_Mae\",\n",
    "    \"QE_I15\": \"UF_Ensino_Medio\",\n",
    "    \"QE_I05\": \"Renda\",\n",
    "    \"QE_I06\": \"Situacao_Financeira\",\n",
    "    \"QE_I03\": \"Moradia\",\n",
    "    \"QE_I04\": \"Qtde_Pessoas\",\n",
    "    \"QE_I12\": \"Politica_de_Ingresso\",\n",
    "    \"QE_I07\": \"Situacao_Trabalho\",\n",
    "    \"QE_I09\": \"Bolsa_de_Estudos\",\n",
    "    \"QE_I19\": \"Qtde_Livros\",\n",
    "    \"QE_I20\": \"Horas_de_Estudo\",\n",
    "    \"QE_I22\": \"Condicao_Salas\",\n",
    "    \"QE_I26\": \"Condicao_Praticas\",\n",
    "    \"QE_I41\": \"Disponibilidade_Professores\",\n",
    "    \"QE_I42\": \"Dominio_Professores\",\n",
    "    \"QE_I52\": \"Plano_de_Ensino\",\n",
    "    \"QE_I54\": \"Formacao\"\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "110807ac-d3b7-49c3-ae9f-e74ce3444e65",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11.rename(columns = columns_dict, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "4a2b29ba-c252-4f62-bb04-93e0057fb3e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11 = dados_11.reindex(columns=[\"Ano\",\"Codigo_do_Curso\",\"Curso\",\"UF\",\"UF_Ensino_Medio\",\"Regiao\",\"Cat_Administrativa\",\"Cat_Org_Academica\",\"Turno\",\n",
    "                                     \"Tipo_Presenca\",\"Nota_FG\",\"Nota_CE\",\"Conclusao_Ensino_Medio\",\"Inicio_Graduacao\",\"Anos_Preparacao\",\"Anos_Graduacao\",\"Tipo_Escola_EM\",\"Modalidade_EM\",\n",
    "                                     \"Idade\",\"Sexo\",\"Cor_Raca\",\"Estado_Civil\",\"Nl_Pai\",\"Nl_Mae\",\"Renda\",\"Situacao_Financeira\",\"Moradia\",\"Qtde_Pessoas\",\n",
    "                                     \"Politica_de_Ingresso\",\"Situacao_Trabalho\",\"Bolsa_de_Estudos\",\"Qtde_Livros\",\"Horas_de_Estudo\",\"Disponibilidade_Professores\",\n",
    "                                     \"Dominio_Professores\",\"Condicao_Salas\",\"Condicao_Praticas\",\"Plano_de_Ensino\",\"Formacao\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ba26ce97-a1a7-4fd5-a273-7ee8f5dc768f",
   "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>Ano</th>\n",
       "      <th>Codigo_do_Curso</th>\n",
       "      <th>Curso</th>\n",
       "      <th>UF</th>\n",
       "      <th>UF_Ensino_Medio</th>\n",
       "      <th>Regiao</th>\n",
       "      <th>Cat_Administrativa</th>\n",
       "      <th>Cat_Org_Academica</th>\n",
       "      <th>Turno</th>\n",
       "      <th>Tipo_Presenca</th>\n",
       "      <th>...</th>\n",
       "      <th>Situacao_Trabalho</th>\n",
       "      <th>Bolsa_de_Estudos</th>\n",
       "      <th>Qtde_Livros</th>\n",
       "      <th>Horas_de_Estudo</th>\n",
       "      <th>Disponibilidade_Professores</th>\n",
       "      <th>Dominio_Professores</th>\n",
       "      <th>Condicao_Salas</th>\n",
       "      <th>Condicao_Praticas</th>\n",
       "      <th>Plano_de_Ensino</th>\n",
       "      <th>Formacao</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2011</td>\n",
       "      <td>36</td>\n",
       "      <td>BCC</td>\n",
       "      <td>MT</td>\n",
       "      <td>nan</td>\n",
       "      <td>CO</td>\n",
       "      <td>Pública Federal</td>\n",
       "      <td>Universidade</td>\n",
       "      <td>Integral</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>Não</td>\n",
       "      <td>nan</td>\n",
       "      <td>1 ou 2</td>\n",
       "      <td>+12</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Não sei responder</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2011</td>\n",
       "      <td>36</td>\n",
       "      <td>BCC</td>\n",
       "      <td>MT</td>\n",
       "      <td>MT</td>\n",
       "      <td>CO</td>\n",
       "      <td>Pública Federal</td>\n",
       "      <td>Universidade</td>\n",
       "      <td>Integral</td>\n",
       "      <td>222</td>\n",
       "      <td>...</td>\n",
       "      <td>21 a 39 horas</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>1 ou 2</td>\n",
       "      <td>nan</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Muito pouco</td>\n",
       "      <td>Fraca</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011</td>\n",
       "      <td>36</td>\n",
       "      <td>BCC</td>\n",
       "      <td>MT</td>\n",
       "      <td>nan</td>\n",
       "      <td>CO</td>\n",
       "      <td>Pública Federal</td>\n",
       "      <td>Universidade</td>\n",
       "      <td>Integral</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>Não</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>nan</td>\n",
       "      <td>4-7</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Muito pouco</td>\n",
       "      <td>Muito fraca</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2011</td>\n",
       "      <td>36</td>\n",
       "      <td>BCC</td>\n",
       "      <td>MT</td>\n",
       "      <td>nan</td>\n",
       "      <td>CO</td>\n",
       "      <td>Pública Federal</td>\n",
       "      <td>Universidade</td>\n",
       "      <td>Integral</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>Até 20 horas</td>\n",
       "      <td>nan</td>\n",
       "      <td>nan</td>\n",
       "      <td>1-3</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Nenhuma</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Parcialmente</td>\n",
       "      <td>Fraca</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2011</td>\n",
       "      <td>36</td>\n",
       "      <td>BCC</td>\n",
       "      <td>MT</td>\n",
       "      <td>MT</td>\n",
       "      <td>CO</td>\n",
       "      <td>Pública Federal</td>\n",
       "      <td>Universidade</td>\n",
       "      <td>Integral</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>Não</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>nan</td>\n",
       "      <td>Nenhum</td>\n",
       "      <td>Não sei responder</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Parcialmente</td>\n",
       "      <td>Muito boa</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",
       "      <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",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41018</th>\n",
       "      <td>2011</td>\n",
       "      <td>5000884</td>\n",
       "      <td>ADS</td>\n",
       "      <td>SP</td>\n",
       "      <td>SP</td>\n",
       "      <td>SE</td>\n",
       "      <td>Pública Estadual</td>\n",
       "      <td>Faculdade</td>\n",
       "      <td>Noturno</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>Não</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>3-5</td>\n",
       "      <td>8-12</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Amplamente</td>\n",
       "      <td>Muito boa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41019</th>\n",
       "      <td>2011</td>\n",
       "      <td>5000884</td>\n",
       "      <td>ADS</td>\n",
       "      <td>SP</td>\n",
       "      <td>SP</td>\n",
       "      <td>SE</td>\n",
       "      <td>Pública Estadual</td>\n",
       "      <td>Faculdade</td>\n",
       "      <td>Matutino</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>40 horas ou mais</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>3-5</td>\n",
       "      <td>8-12</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Nenhuma</td>\n",
       "      <td>Parcialmente</td>\n",
       "      <td>Boa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41020</th>\n",
       "      <td>2011</td>\n",
       "      <td>5000884</td>\n",
       "      <td>ADS</td>\n",
       "      <td>SP</td>\n",
       "      <td>SP</td>\n",
       "      <td>SE</td>\n",
       "      <td>Pública Estadual</td>\n",
       "      <td>Faculdade</td>\n",
       "      <td>Matutino</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>40 horas ou mais</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>+8</td>\n",
       "      <td>1-3</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Amplamente</td>\n",
       "      <td>Boa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41021</th>\n",
       "      <td>2011</td>\n",
       "      <td>5000884</td>\n",
       "      <td>ADS</td>\n",
       "      <td>SP</td>\n",
       "      <td>SP</td>\n",
       "      <td>SE</td>\n",
       "      <td>Pública Estadual</td>\n",
       "      <td>Faculdade</td>\n",
       "      <td>Matutino</td>\n",
       "      <td>555</td>\n",
       "      <td>...</td>\n",
       "      <td>40 horas ou mais</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>1 ou 2</td>\n",
       "      <td>1-3</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Nenhuma</td>\n",
       "      <td>Amplamente</td>\n",
       "      <td>Regular</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41022</th>\n",
       "      <td>2011</td>\n",
       "      <td>5000884</td>\n",
       "      <td>ADS</td>\n",
       "      <td>SP</td>\n",
       "      <td>SP</td>\n",
       "      <td>SE</td>\n",
       "      <td>Pública Estadual</td>\n",
       "      <td>Faculdade</td>\n",
       "      <td>Matutino</td>\n",
       "      <td>222</td>\n",
       "      <td>...</td>\n",
       "      <td>Não</td>\n",
       "      <td>Não se aplica</td>\n",
       "      <td>1 ou 2</td>\n",
       "      <td>Nenhum</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Alguns</td>\n",
       "      <td>Algumas</td>\n",
       "      <td>Maior parte</td>\n",
       "      <td>Parcialmente</td>\n",
       "      <td>Boa</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>41023 rows × 39 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        Ano  Codigo_do_Curso Curso  UF UF_Ensino_Medio Regiao  \\\n",
       "0      2011               36   BCC  MT             nan     CO   \n",
       "1      2011               36   BCC  MT              MT     CO   \n",
       "2      2011               36   BCC  MT             nan     CO   \n",
       "3      2011               36   BCC  MT             nan     CO   \n",
       "4      2011               36   BCC  MT              MT     CO   \n",
       "...     ...              ...   ...  ..             ...    ...   \n",
       "41018  2011          5000884   ADS  SP              SP     SE   \n",
       "41019  2011          5000884   ADS  SP              SP     SE   \n",
       "41020  2011          5000884   ADS  SP              SP     SE   \n",
       "41021  2011          5000884   ADS  SP              SP     SE   \n",
       "41022  2011          5000884   ADS  SP              SP     SE   \n",
       "\n",
       "      Cat_Administrativa Cat_Org_Academica     Turno  Tipo_Presenca  ...  \\\n",
       "0        Pública Federal      Universidade  Integral            555  ...   \n",
       "1        Pública Federal      Universidade  Integral            222  ...   \n",
       "2        Pública Federal      Universidade  Integral            555  ...   \n",
       "3        Pública Federal      Universidade  Integral            555  ...   \n",
       "4        Pública Federal      Universidade  Integral            555  ...   \n",
       "...                  ...               ...       ...            ...  ...   \n",
       "41018   Pública Estadual         Faculdade   Noturno            555  ...   \n",
       "41019   Pública Estadual         Faculdade  Matutino            555  ...   \n",
       "41020   Pública Estadual         Faculdade  Matutino            555  ...   \n",
       "41021   Pública Estadual         Faculdade  Matutino            555  ...   \n",
       "41022   Pública Estadual         Faculdade  Matutino            222  ...   \n",
       "\n",
       "       Situacao_Trabalho  Bolsa_de_Estudos  Qtde_Livros  Horas_de_Estudo  \\\n",
       "0                    Não               nan       1 ou 2              +12   \n",
       "1          21 a 39 horas     Não se aplica       1 ou 2              nan   \n",
       "2                    Não     Não se aplica          nan              4-7   \n",
       "3           Até 20 horas               nan          nan              1-3   \n",
       "4                    Não     Não se aplica          nan           Nenhum   \n",
       "...                  ...               ...          ...              ...   \n",
       "41018                Não     Não se aplica          3-5             8-12   \n",
       "41019   40 horas ou mais     Não se aplica          3-5             8-12   \n",
       "41020   40 horas ou mais     Não se aplica           +8              1-3   \n",
       "41021   40 horas ou mais     Não se aplica       1 ou 2              1-3   \n",
       "41022                Não     Não se aplica       1 ou 2           Nenhum   \n",
       "\n",
       "       Disponibilidade_Professores  Dominio_Professores     Condicao_Salas  \\\n",
       "0                Não sei responder    Não sei responder  Não sei responder   \n",
       "1                           Alguns               Alguns            Algumas   \n",
       "2                           Alguns               Alguns            Algumas   \n",
       "3                           Alguns               Alguns            Nenhuma   \n",
       "4                Não sei responder          Maior parte            Algumas   \n",
       "...                            ...                  ...                ...   \n",
       "41018                       Alguns          Maior parte        Maior parte   \n",
       "41019                       Alguns               Alguns            Algumas   \n",
       "41020                       Alguns          Maior parte        Maior parte   \n",
       "41021                  Maior parte          Maior parte            Algumas   \n",
       "41022                       Alguns               Alguns            Algumas   \n",
       "\n",
       "       Condicao_Praticas    Plano_de_Ensino           Formacao  \n",
       "0      Não sei responder  Não sei responder  Não sei responder  \n",
       "1                Algumas        Muito pouco              Fraca  \n",
       "2                Algumas        Muito pouco        Muito fraca  \n",
       "3                Algumas       Parcialmente              Fraca  \n",
       "4                Algumas       Parcialmente          Muito boa  \n",
       "...                  ...                ...                ...  \n",
       "41018        Maior parte         Amplamente          Muito boa  \n",
       "41019            Nenhuma       Parcialmente                Boa  \n",
       "41020        Maior parte         Amplamente                Boa  \n",
       "41021            Nenhuma         Amplamente            Regular  \n",
       "41022        Maior parte       Parcialmente                Boa  \n",
       "\n",
       "[41023 rows x 39 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dados_11"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1818a832-e09a-43b0-ae3c-435b3b96f36b",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11 = dados_11[dados_11[\"Codigo_do_Curso\"] != 94260]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "75691cf8-3aa6-4a8b-855d-13b3b7340791",
   "metadata": {},
   "outputs": [],
   "source": [
    "dados_11.to_csv('./{}.csv'.format('pre-process_2011'),index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d33ee41e-5207-4ff1-aebe-b7746a9477c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Gráfico do número de alunos de cada curso"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "962a1aa9-6c64-47bb-8a54-85f28598d716",
   "metadata": {},
   "outputs": [],
   "source": [
    "count_bcc = dados_11[dados_11['Curso'] == 'BCC']['Curso'].count()\n",
    "count_ads = dados_11[dados_11['Curso'] == 'ADS']['Curso'].count()\n",
    "count_gti = dados_11[dados_11['Curso'] == 'GTI']['Curso'].count()\n",
    "count_ec = dados_11[dados_11['Curso'] == 'EC']['Curso'].count()\n",
    "count_lcc = dados_11[dados_11['Curso'] == 'LCC']['Curso'].count()\n",
    "count_rc = dados_11[dados_11['Curso'] == 'RC']['Curso'].count()\n",
    "count_si = dados_11[dados_11['Curso'] == 'SI']['Curso'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e7c22705-62cc-4f11-a9c8-df5d9b5722c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "count_array = [count_bcc,count_ads, count_gti, count_ec, \n",
    "               count_lcc, count_rc, count_si]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c95b6bba-d3a5-46d3-94aa-dd79e1e7d048",
   "metadata": {},
   "outputs": [],
   "source": [
    "count_courses = pd.DataFrame({'Número de alunos': count_array}, index=['BCC', 'ADS', 'GTI', 'EC', 'LCC', 'RC', 'SI'])\n",
    "\n",
    "count_courses.plot.bar();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7d8a2500-5a27-44f8-b552-1bd859104fba",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}