testthat/old/tjrs_revisao_v2.R

In filipezabala/jurimetrics: General Purpose Tools for Jurimetrics

# restart R
# .rs.restartR()

# pacotes
library(tidyverse)  # varios
library(readxl)     # read_xlsx
library(lubridate)  # wday
library(qcc)        # pareto.chart

# functions
devtools::load_all()
source('~/Dropbox/Jurimetria/codigos/git/jurimetrics/fits.R')

# set strings as factors to false
options(stringsAsFactors = FALSE)

# diretorio
setwd('~/Desktop/wbs/arquivosLiteExcel/')            #  Mac
# setwd('~/Área de Trabalho/wbs/arquivosExcel')   # Linux

# lendo bancos de dados
ini <- Sys.time()
data <- vector('list')
for(i in 2000:2017){
  df <- read_xlsx(paste0('tjrs',i,'.xlsx'))
  temp <- substr(df$`Data de Julgamento`,3,12)
  data[[paste0(i)]] <- as_tibble(as.Date(temp,  format = '%d/%m/%Y'))
  names(data[[paste0(i)]]) <- 'judgmentDate'
  print(paste0(i))
}
gc()
rm(df)
Sys.time() - ini # Time difference of 7.998097 mins
 

# criando colunas de semana e mês
ini <- Sys.time()
data <- lapply(data, mutate, weekDay = lubridate::wday(judgmentDate, label = T))
data <- lapply(data, mutate, yearMonth = as.Date(paste0(substr(judgmentDate,1,7),'-01'), 
                                                  format = '%Y-%m-%d'))
# empilhando
data <- do.call(rbind, data)
Sys.time() - ini # Time difference of 27.99515 secs

# agrupando e contando o número de processos por dia
count_day <- data %>% 
  group_by(judgmentDate) %>%
  summarize(count = n())
write_csv(count_day, '~/Dropbox/Jurimetria/codigos/git/jurimetrics/count_day.csv')
  
# agrupando e contando o número de processos por dia a semana
count_week_day <- data %>% 
  group_by(weekDay) %>%
  summarize(count = n())
write_csv(count_week_day, '~/Dropbox/Jurimetria/codigos/git/jurimetrics/count_week_day.csv')
pareto.chart(table(data$weekDay))

# agrupando e contando o número de processos por mês
count_year_month <- data %>% 
  group_by(yearMonth) %>%
  summarize(count = n())
write_csv(count_year_month, '~/Dropbox/Jurimetria/codigos/git/jurimetrics/count_year_month.csv')
ggplot(count_year_month, aes(yearMonth, count)) +
  geom_line()

# projetando
ini <- Sys.time()
y <- ts(count_year_month$count, start = c(2000,1), frequency = 12)
fits(y, steps = 48, lim = F, graf = T, PI = F)
Sys.time()-ini   # Time difference of 6.028008 mins, PI = T