dds |> gg_season(ibcr_ce) |> plotly_build()
dds |> gg_subseries(ibcr_ce) |> plotly_build()
dds |>
model(classical_decomposition(ibcr_ce,
type = "additive")) |>
components() |>
autoplot() |>
labs(title = "Decomposição Adivita por Média Móvel do IBCR Cearense")
## [[1]]
##
## $title
## [1] "Decomposição Adivita por Média Móvel do IBCR Cearense"
##
## attr(,"class")
## [1] "labels"
dds |>
features(ibcr_ce,c(unitroot_kpss, unitroot_pp))
## # A tibble: 1 × 4
## kpss_stat kpss_pvalue pp_stat pp_pvalue
## <dbl> <dbl> <dbl> <dbl>
## 1 3.60 0.01 -2.81 0.0600
ce_models <-dds |>
model(ce_auto = ARIMA(ibcr_ce ~ 0 + pdq(2,1,1) + PDQ(0,1,2, period = "1 year")))
glance(ce_models) |>
arrange(AICc) |>
select(.model:BIC)
## # A tibble: 1 × 6
## .model sigma2 log_lik AIC AICc BIC
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 ce_auto 8.22 -539. 1091. 1091. 1111.
ggplotly(ce_models |>
dplyr::select(ce_auto) |>
fitted() |>
ggplot() +
geom_line(aes(x=dt, y=.fitted),
alpha = .66, size= .8,
linetype="twodash") +
geom_line(aes(x=dt, y=dds$ibcr_ce),
color = 'seagreen4',
alpha = .75, size= .8))
ce_models |>
dplyr::select(ce_auto) |>
fitted() |>
ggplot() +
geom_col(aes(x=dt, y=(.fitted - dds$ibcr_ce)), color='seagreen4') +
labs(title = 'Erros entre Estimado e Valor Real da Série Cearense')
ce_models |>
dplyr::select(ce_auto) |>
gg_tsresiduals()
prev_ce <- ce_models |> dplyr::select(ce_auto) |> fabletools::forecast(h=12)
prev_ce |> autoplot(dds)
time_prev <- seq(as.Date("2018-01-01"),
as.Date("2023-01-01"),
by = "1 month") |> yearmonth()
ggplot() +
geom_line(aes(
x=time_prev,
y= c(dds$ibcr_ce[dds$dt >= yearmonth('2018 Jan')],
prev_ce$.mean)),
color='tomato3',
size=.8) +
geom_line(aes(x=prev_ce$dt,
y=prev_ce$.mean),
size=.8) +
labs(title ='Previsão do IBCR Cearense',
subtitle = "ARIMA(2,1,1)(0,1,2)(12)",
x= 'Tempo',
y='Índice')
