Regressão Romântica - Speed Dating
André Felipe
Neste problema, utilizaremos dados românticos descritos e disponíveis aqui: https://github.com/nazareno/ciencia-de-dados-1/tree/master/5-regressao/speed-dating
Especialmente, atente para uma coluna chamada dec, que diz se houve match entre os dois participantes do encontro – isso é: ambos disseram que gostariam de se encontrar novamente depois: https://github.com/nazareno/ciencia-de-dados-1/blob/master/5-regressao/speed-dating/speed-dating2.csv
Sua missão é utilizar regressão logística em um conjunto de variáveis explicativas que você escolher (com no mínimo 4 variáveis) para responder o seguinte com esses dados em um RMarkdown:
Que fatores nos dados têm efeito re;evante na chance do casal ter um match? Descreva se os efeitos são positivos ou negativos e sua magnitude.
Lembre que temos apenas uma amostra de encontros.
Lembre de fazer um descritivo das variáveis antes, e de escrever o relatório de maneira que alguém que saiba sobre regressão mas não sabe nada sobre os dados entenda.
Registered S3 method overwritten by 'dplyr':
method from
print.rowwise_df
[30m── [1mAttaching packages[22m ─────────────────────────────────────────────────────────── tidyverse 1.2.1 ──[39m
[30m[32m✔[30m [34mggplot2[30m 3.2.0 [32m✔[30m [34mpurrr [30m 0.3.2
[32m✔[30m [34mtibble [30m 2.1.3 [32m✔[30m [34mdplyr [30m 0.8.3
[32m✔[30m [34mtidyr [30m 0.8.3 [32m✔[30m [34mstringr[30m 1.4.0
[32m✔[30m [34mreadr [30m 1.3.1 [32m✔[30m [34mforcats[30m 0.4.0[39m
[30m── [1mConflicts[22m ────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
[31m✖[30m [34mdplyr[30m::[32mfilter()[30m masks [34mstats[30m::filter()
[31m✖[30m [34mdplyr[30m::[32mlag()[30m masks [34mstats[30m::lag()[39m
Attaching package: ‘modelr’
The following object is masked from ‘package:broom’:
bootstrap
here() starts at /cloud/project/5-regressao
Registered S3 method overwritten by 'GGally':
method from
+.gg ggplot2
Attaching package: ‘GGally’
The following object is masked from ‘package:dplyr’:
nasa
Classes and Methods for R developed in the
Political Science Computational Laboratory
Department of Political Science
Stanford University
Simon Jackman
hurdle and zeroinfl functions by Achim ZeileisDados dos speed datings
dados = read_csv(here::here("speed-dating/speed-dating2.csv"))Parsed with column specification:
cols(
.default = col_double(),
field = [31mcol_character()[39m,
from = [31mcol_character()[39m,
career = [31mcol_character()[39m,
attr3_s = [33mcol_logical()[39m,
sinc3_s = [33mcol_logical()[39m,
intel3_s = [33mcol_logical()[39m,
fun3_s = [33mcol_logical()[39m,
amb3_s = [33mcol_logical()[39m,
dec = [31mcol_character()[39m
)
See spec(...) for full column specifications.
10220 parsing failures.
row col expected actual file
1847 attr3_s 1/0/T/F/TRUE/FALSE 8.00 '/cloud/project/5-regressao/speed-dating/speed-dating2.csv'
1847 sinc3_s 1/0/T/F/TRUE/FALSE 10.00 '/cloud/project/5-regressao/speed-dating/speed-dating2.csv'
1847 intel3_s 1/0/T/F/TRUE/FALSE 9.00 '/cloud/project/5-regressao/speed-dating/speed-dating2.csv'
1847 fun3_s 1/0/T/F/TRUE/FALSE 10 '/cloud/project/5-regressao/speed-dating/speed-dating2.csv'
1847 amb3_s 1/0/T/F/TRUE/FALSE 10 '/cloud/project/5-regressao/speed-dating/speed-dating2.csv'
.... ........ .................. ...... ...........................................................
See problems(...) for more details.dados = dados %>%
filter(!is.na(attr), !is.na(fun), !is.na(intel), !is.na(sinc), !is.na(amb), !is.na(shar), !is.na(prob), !is.na(like)) %>%
mutate(dec = case_when(.$dec == "no" ~ 0,
.$dec == "yes" ~ 1))
glimpse(dados)Observations: 4,101
Variables: 44
$ iid [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3,…
$ gender [3m[38;5;246m<dbl>[39m[23m 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…
$ order [3m[38;5;246m<dbl>[39m[23m 4, 3, 5, 7, 6, 1, 2, 8, 9, 10, 9, 6, 1, 3, 2, 7, 8, 4, 5, 6, 5, 2, 7, 3, 4, 10, …
$ pid [3m[38;5;246m<dbl>[39m[23m 11, 12, 14, 15, 16, 17, 18, 19, 20, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 11, …
$ int_corr [3m[38;5;246m<dbl>[39m[23m 0.14, 0.54, 0.61, 0.21, 0.25, 0.34, 0.50, 0.28, -0.36, 0.29, 0.18, 0.10, -0.21, …
$ samerace [3m[38;5;246m<dbl>[39m[23m 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,…
$ age_o [3m[38;5;246m<dbl>[39m[23m 27, 22, 23, 24, 25, 30, 27, 28, 24, 27, 22, 22, 23, 24, 25, 30, 27, 28, 24, 27, …
$ age [3m[38;5;246m<dbl>[39m[23m 21, 21, 21, 21, 21, 21, 21, 21, 21, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 25, …
$ field [3m[38;5;246m<chr>[39m[23m "Law", "Law", "Law", "Law", "Law", "Law", "Law", "Law", "Law", "law", "law", "la…
$ race [3m[38;5;246m<dbl>[39m[23m 4, 4, 4, 4, 4, 4, 4, 4, 4, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,…
$ from [3m[38;5;246m<chr>[39m[23m "Chicago", "Chicago", "Chicago", "Chicago", "Chicago", "Chicago", "Chicago", "Ch…
$ career [3m[38;5;246m<chr>[39m[23m "lawyer", "lawyer", "lawyer", "lawyer", "lawyer", "lawyer", "lawyer", "lawyer", …
$ sports [3m[38;5;246m<dbl>[39m[23m 9, 9, 9, 9, 9, 9, 9, 9, 9, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,…
$ tvsports [3m[38;5;246m<dbl>[39m[23m 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 8, 8, 8, 8, 8, 8, 8, 8,…
$ exercise [3m[38;5;246m<dbl>[39m[23m 8, 8, 8, 8, 8, 8, 8, 8, 8, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,…
$ dining [3m[38;5;246m<dbl>[39m[23m 9, 9, 9, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 8, 8, 8, 8, 8…
$ museums [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 5, 5, 5, 5, 5, 5, 5, 5,…
$ art [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 5, 5, 5, 5, 5, 5, 5,…
$ hiking [3m[38;5;246m<dbl>[39m[23m 5, 5, 5, 5, 5, 5, 5, 5, 5, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 8, 8, 8, 8, 8, 8, 8, 8,…
$ gaming [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,…
$ clubbing [3m[38;5;246m<dbl>[39m[23m 5, 5, 5, 5, 5, 5, 5, 5, 5, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 5, 5, 5, 5, 5, 5, 5, 5,…
$ reading [3m[38;5;246m<dbl>[39m[23m 6, 6, 6, 6, 6, 6, 6, 6, 6, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 7, 7, 7, 7, 7…
$ tv [3m[38;5;246m<dbl>[39m[23m 9, 9, 9, 9, 9, 9, 9, 9, 9, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 8, 8, 8, 8, 8, 8, 8, 8,…
$ theater [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 7, 7, 7, 7, 7, 7, 7, 7,…
$ movies [3m[38;5;246m<dbl>[39m[23m 10, 10, 10, 10, 10, 10, 10, 10, 10, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 7, 7, 7, 7, 7,…
$ concerts [3m[38;5;246m<dbl>[39m[23m 10, 10, 10, 10, 10, 10, 10, 10, 10, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,…
$ music [3m[38;5;246m<dbl>[39m[23m 9, 9, 9, 9, 9, 9, 9, 9, 9, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 5, 5, 5, 5, 5, 5, 5, 5,…
$ shopping [3m[38;5;246m<dbl>[39m[23m 8, 8, 8, 8, 8, 8, 8, 8, 8, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 8, 8, 8, 8, 8, 8, 8, 8,…
$ yoga [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 7, 7, 7, 7, 7, 7, 7, 7,…
$ attr [3m[38;5;246m<dbl>[39m[23m 6, 7, 7, 5, 4, 7, 4, 7, 5, 5, 8, 5, 7, 6, 8, 7, 5, 7, 6, 7, 9, 7, 9, 8, 7, 9, 8,…
$ sinc [3m[38;5;246m<dbl>[39m[23m 9, 8, 6, 6, 9, 6, 9, 6, 6, 7, 5, 8, 9, 8, 7, 5, 8, 6, 7, 9, 7, 9, 7, 9, 9, 9, 7,…
$ intel [3m[38;5;246m<dbl>[39m[23m 7, 7, 8, 7, 7, 7, 7, 8, 6, 8, 6, 9, 7, 7, 8, 9, 7, 8, 8, 10, 9, 9, 9, 10, 9, 9, …
$ fun [3m[38;5;246m<dbl>[39m[23m 7, 8, 7, 7, 4, 4, 6, 9, 8, 4, 6, 6, 6, 9, 3, 6, 5, 9, 7, 7, 8, 7, 7, 7, 8, 9, 7,…
$ amb [3m[38;5;246m<dbl>[39m[23m 6, 5, 6, 6, 6, 6, 5, 8, 10, 6, 9, 3, 5, 7, 6, 7, 9, 4, 9, 8, 9, 9, 9, 7, 9, 9, 9…
$ shar [3m[38;5;246m<dbl>[39m[23m 5, 6, 8, 6, 4, 7, 6, 8, 8, 3, 6, 4, 7, 8, 2, 9, 5, 5, 8, 9, 7, 7, 7, 9, 7, 9, 7,…
$ like [3m[38;5;246m<dbl>[39m[23m 7, 7, 7, 6, 6, 6, 6, 7, 6, 6, 7, 6, 7, 8, 6, 8, 5, 5, 8, 8, 8, 8, 8, 8, 8, 9, 8,…
$ prob [3m[38;5;246m<dbl>[39m[23m 6, 5, 6, 6, 5, 5, 7, 7, 6, 4, 3, 7, 8, 6, 5, 7, 6, 6, 7, 7, 7, 7, 7, 7, 7, 7, 7,…
$ match_es [3m[38;5;246m<dbl>[39m[23m 4, 4, 4, 4, 4, 4, 4, 4, 4, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, NA, NA, NA, NA, NA, NA,…
$ attr3_s [3m[38;5;246m<lgl>[39m[23m NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
$ sinc3_s [3m[38;5;246m<lgl>[39m[23m NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
$ intel3_s [3m[38;5;246m<lgl>[39m[23m NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
$ fun3_s [3m[38;5;246m<lgl>[39m[23m NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
$ amb3_s [3m[38;5;246m<lgl>[39m[23m NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
$ dec [3m[38;5;246m<dbl>[39m[23m 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0,…Olhada rápida nos dados
skimr::skim(dados)Skim summary statistics
n obs: 4101
n variables: 44
── Variable type:character ─────────────────────────────────────────────────────────────────────────
variable missing complete n min max empty n_unique
career 38 4063 4101 2 77 0 215
field 13 4088 4101 3 51 0 145
from 28 4073 4101 2 58 0 171
── Variable type:logical ───────────────────────────────────────────────────────────────────────────
variable missing complete n mean count
amb3_s 4101 0 4101 NaN 4101
attr3_s 4101 0 4101 NaN 4101
fun3_s 4101 0 4101 NaN 4101
intel3_s 4101 0 4101 NaN 4101
sinc3_s 4101 0 4101 NaN 4101
── Variable type:numeric ───────────────────────────────────────────────────────────────────────────
variable missing complete n mean sd p0 p25 p50 p75 p100 hist
age 37 4064 4101 25.7 3.23 18 23 25 28 38 ▁▇▆▇▃▁▁▁
age_o 43 4058 4101 25.72 3.32 18 23 25 28 39 ▁▇▆▇▃▁▁▁
amb 0 4101 4101 6.68 1.84 0 5.5 7 8 10 ▁▁▁▇▇▇▇▆
art 28 4073 4101 6.59 2.33 0 5 7 8 10 ▁▁▃▇▃▆▇▇
attr 0 4101 4101 6.05 1.94 0 5 6 7 10 ▁▁▂▇▆▆▅▃
clubbing 28 4073 4101 5.72 2.46 0 4 6 8 10 ▃▂▃▇▆▆▇▅
concerts 28 4073 4101 6.81 2.14 0 6 7 8 10 ▁▁▂▅▆▆▆▇
dec 0 4101 4101 0.42 0.49 0 0 0 1 1 ▇▁▁▁▁▁▁▆
dining 28 4073 4101 7.69 1.77 1 7 8 9 10 ▁▁▁▂▂▅▅▇
exercise 28 4073 4101 6.13 2.35 1 5 6 8 10 ▅▃▃▆▇▇▇▇
fun 0 4101 4101 6.27 1.99 0 5 6 8 10 ▁▁▂▇▇▇▆▅
gaming 28 4073 4101 4.04 2.65 0 2 4 6 14 ▆▇▆▆▂▁▁▁
gender 0 4101 4101 0.5 0.5 0 0 0 1 1 ▇▁▁▁▁▁▁▇
hiking 28 4073 4101 5.66 2.54 0 4 6 8 10 ▂▃▅▇▆▅▆▆
iid 0 4101 4101 273.48 184.73 1 87 271 433 552 ▇▇▁▆▃▅▃▇
int_corr 52 4049 4101 0.19 0.31 -0.73 -0.03 0.21 0.43 0.9 ▁▂▅▆▇▇▅▁
intel 0 4101 4101 7.25 1.59 0 6 7 8 10 ▁▁▁▃▅▇▇▆
like 0 4101 4101 6.03 1.86 0 5 6 7 10 ▁▁▂▇▇▇▅▂
match_es 348 3753 4101 3.1 2.4 0 2 3 4 10 ▇▇▆▆▁▁▁▂
movies 28 4073 4101 7.94 1.72 0 7 8 9 10 ▁▁▁▁▂▃▅▇
museums 28 4073 4101 6.89 2.11 0 6 7 8 10 ▁▁▂▅▃▇▆▇
music 28 4073 4101 7.8 1.88 1 7 8 9 10 ▁▁▁▂▂▃▃▇
order 0 4101 4101 9.09 5.61 1 4 8 13 22 ▇▇▅▆▅▃▃▂
pid 8 4093 4101 273.74 184.24 1 88 273 433 552 ▇▇▁▆▃▅▃▇
prob 0 4101 4101 5.04 2.18 0 4 5 7 10 ▂▂▂▇▃▃▂▁
race 13 4088 4101 2.75 1.23 1 2 2 4 6 ▁▇▁▁▃▁▁▁
reading 28 4073 4101 7.65 2.06 1 7 8 9 13 ▁▁▁▅▃▇▁▁
samerace 0 4101 4101 0.4 0.49 0 0 0 1 1 ▇▁▁▁▁▁▁▆
shar 0 4101 4101 5.3 2.16 0 4 5 7 10 ▁▂▂▇▅▃▂▂
shopping 28 4073 4101 5.51 2.59 1 4 6 7 10 ▇▃▅▆▆▆▅▆
sinc 0 4101 4101 7.03 1.81 0 6 7 8 10 ▁▁▁▅▆▇▇▆
sports 28 4073 4101 6.37 2.53 1 5 7 8 10 ▃▂▃▃▅▅▆▇
theater 28 4073 4101 6.69 2.28 0 5 7 8 10 ▁▁▂▆▃▆▅▇
tv 28 4073 4101 5.32 2.47 1 3 6 7 10 ▇▃▃▆▇▆▆▃
tvsports 28 4073 4101 4.55 2.81 1 2 4 7 10 ▇▃▂▂▂▂▂▃
yoga 28 4073 4101 4.17 2.72 0 2 3 6 10 ▇▆▆▆▃▃▂▃Após essa análise rápida, vamos analisar melhor alguns dos dados no próximo tópico.
Histogramas
Faremos a análise das informações que o participante dá uma nota de 1 a 10 no formulário que recebe. Essas informações são as seguintes:
- attr : quão atraente p1 achou p2
- sinc : quão sincero p1 achou p2
- intel : quão inteligente p1 achou p2
- fun : quão divertido p1 achou p2
- amb : quão ambicioso p1 achou p2
- shar : quanto p1 achou que compartilha interesses e hobbies com p2
- like : no geral, quanto p1 gostou de p2
- prob : que probabilidade p1 acha que p2 tem de querer se encontrar novamente com p
As quantidades para cada resposta foram:
dados %>%
ggplot(aes(attr)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(sinc)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(intel)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(fun)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(amb)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(shar)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(like)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
dados %>%
ggplot(aes(prob)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 10, 1))
Com isso, temos material para inferir que talvez essas variáveis se relacionem e possam influenciar no match, visto que essas distribuições das respostas se aproximam de uma distribuição normal. Veremos também o número de decisões “não” (0 no gráfico) e “sim” (1 no gráfico)
# decisoes (0 = no, 1 = yes)
dados %>%
ggplot(aes(dec)) +
geom_bar() +
scale_x_continuous(breaks=seq(0, 1, by = 1))
Visto isso, parece que há alguma relação entre as variáveis escolhidas e a decisão de se encontrar novamente. Já podemos passar para a modelagem.
Modelando a regressão logística multivariada
modelo1 <- glm(dec ~ attr + fun + intel + sinc + amb + shar + like + prob,
data = dados,
family = "binomial")
tidy(modelo1, conf.int = TRUE)tidy(modelo1, conf.int = TRUE, exponentiate = TRUE)
glance(modelo1)pR2(modelo1) llh llhNull G2 McFadden r2ML r2CU
-1888.4241414 -2793.2364426 1809.6246024 0.3239297 0.3567773 0.4795981 Com isso, podemos obter uma regressão dos speed datings:
dec = 0,43 attr + 0,12 fun - 0,04 intel - 0,23 sinc - 0,21 amb + 0,15 shar + 0,62 like + 0,16 prob - 5,99
Descrevendo se os efeitos são positivos ou negativos e sua magnitude:
- attr : o aumento de 1 unidade de atraente produz um aumento de 0,43 na decisão
- fun : o aumento de 1 unidade de divertido produz um aumento de 0,12 na decisão
- intel : o aumento de 1 unidade de inteligente produz uma redução de 0,04 na decisão
- sinc : o aumento de 1 unidade de sincero produz uma redução de 0,23 na decisão
- amb : o aumento de 1 unidade de ambicioso produz uma redução de 0,21 na decisão
- shar : o aumento de 1 unidade de compartilha interesses e hobbies produz um aumento de 0,15 na decisão
- like : o aumento de 1 unidade de quanto gostou produz um aumento de 0,62 na decisão
- prob : o aumento de 1 unidade de probabilidade de querer se encontrar novamente produz um aumento de 0,16 na decisão
Tudo isso aconteceria caso a regressão fosse linear. Como não é, precisamos EXPONENCIAR.
EXPONENCIANDO:
Como aqui y = exp(b0)exp(b1x1), aumentar em uma unidade x, faz com que y seja multiplicado por exp(b1)
dec = exp(1.54 * attr) * exp(1.13 * fun) * exp(0.95 * intel) * exp(0.79 * sinc) * exp(0.80 * amb) * exp(1.16 * shar) * exp(1.85 * like) * exp(1.18 * prob) * 0.002
Descrevendo se os efeitos são positivos ou negativos e sua magnitude:
- attr : aumentar em uma unidade faz com que y seja multiplicado por 1.54
- fun : aumentar em uma unidade faz com que y seja multiplicado por 1.13
- intel : aumentar em uma unidade faz com que y seja multiplicado por 0.95
- sinc : aumentar em uma unidade faz com que y seja multiplicado por 0.79
- amb : aumentar em uma unidade faz com que y seja multiplicado por 0.80
- shar : aumentar em uma unidade faz com que y seja multiplicado por 1.16
- like : aumentar em uma unidade faz com que y seja multiplicado por 1.85
- prob : aumentar em uma unidade faz com que y seja multiplicado por 1.18
Plotando o gráfico para a regressão utilizando apenas dec e fun:
dados %>% ggplot(aes(fun, dec))+geom_jitter(size=0.5)+geom_smooth(method = "glm",
method.args = list(family = "binomial"),
se = FALSE)
Fit relacionando preferências e gêneros em speed dating
dados %>%
mutate(dec = as.factor(dec),
gender = as.factor(gender))gendermodel = glm(dec ~ gender,
data = dados,
family = "binomial")
tidy(gendermodel, conf.int = TRUE, exponentiate = TRUE)glance(gendermodel)pR2(gendermodel) llh llhNull G2 McFadden r2ML r2CU
-2.760453e+03 -2.793236e+03 6.556702e+01 1.173675e-02 1.586093e-02 2.132106e-02 expectativa_realidade <- augment(gendermodel,
type.predict = "response")
expectativa_realidade %>%
mutate(genderNum = ifelse(gender == "1", 1, 0)) %>%
ggplot(aes(x = gender )) +
geom_count(aes(y = genderNum), alpha = 0.5) +
geom_line(aes(y = .fitted))
expectativa_realidade = expectativa_realidade %>%
mutate(categoria_prevista = ifelse(.fitted > .5, "1", "0"))
table(expectativa_realidade$categoria_prevista, expectativa_realidade$gender)
0 1
0 2068 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