tests/testthat/test_arima_2.R

In exploratory-io/exploratory_func: R package for Exploratory

context("test exp_arima")
test_that("exp_arima with aggregation", {
  data("raw_data", package = "AnomalyDetection")
  raw_data$timestamp <- as.POSIXct(raw_data$timestamp)

  # Create time gap intentionally.
  raw_data <- raw_data %>% filter(as.Date(timestamp) != as.Date("1980-09-26"))

  raw_data <- raw_data %>% rename(`time stamp`=timestamp, `cou nt`=count)
  raw_data$pre_col1 <- runif(nrow(raw_data))
  raw_data$pre_col2 <- runif(nrow(raw_data))

  model_df <- raw_data %>%
    exp_arima(`time stamp`, valueColumn=`cou nt`, periods=10, time_unit = "day", na_fill_type="value", test_mode=F)

  # Test output columns for difference data frame.
  ret <- model_df %>% dplyr::select(difference) %>% tidyr::unnest(difference)
  # With fable 0.2.1, it used to pick a model with seasonality, but with fable 0.3.0 it picks a model without seasonality for some reason.
  # expect_true(all(c('time stamp','cou nt','diff_order','seasonal_diff_order','seasonal_diff_period','p_value') %in% colnames(ret)))
  expect_true(all(c('time stamp','cou nt','diff_order','p_value') %in% colnames(ret)))

  ret <- model_df %>% dplyr::select(residuals) %>% tidyr::unnest(residuals)
  ret <- model_df %>% dplyr::select(residual_acf) %>% tidyr::unnest() %>% rename(Lag=lag, ACF=acf)
  # No valueColumn (row number) case.
  model_df <- raw_data %>%
    exp_arima(`time stamp`, ,periods=10, time_unit = "day", na_fill_type="value", test_mode=F)
  # With more parameters. Revive this test when we support ARIMAX.
  # model_df <- raw_data %>%
  #   exp_arima(`time stamp`, `pre_col1`, `pre_col2`, time_unit = "day", periods = 2, valueColumn = `cou nt`, fun.aggregate = sum, na_fill_type="value", test_mode =TRUE, d = NA, D = NA, max.p = 5, max.q = 5, max.P = 2, max.Q = 2, max.order = 5, max.d = 2, max.D = 1, start.p = 2, start.q = 2, start.P = 1, start.Q = 1, stationary = FALSE, seasonal = TRUE, ic = "aic", allowdrift = TRUE, allowmean = TRUE, lambda = NULL, biasadj = FALSE, test = "kpss", seasonal.test = "ocsb", parallel = FALSE, num.cores = 2)

  expect_true(!is.null(model_df$model))
  # test for glance.
  ret <- model_df %>% glance_rowwise(model)
  ret <- model_df %>% glance_with_ts_metric()
  expect_true(all(c("RMSE","MAE","MAPE (Ratio)","R Squared") %in% names(ret)))
})

test_that("exp_arima with column names including special characters with group_by." , {
  ts <- seq.Date(as.Date("2010-01-01"), as.Date("2010-01-13"), by="day")
  raw_data <- data.frame(timestamp=ts, data=runif(length(ts)), group="A") %>% 
    dplyr::rename(`time stamp !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`=timestamp, `data !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`=data, `group !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`=group) %>%
    dplyr::group_by(`group !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`)
  model_df <- raw_data %>%
    exp_arima(`time stamp !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`, `data !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`, 10, time_unit = "day", funs.aggregate.regressors = c(mean), yearly.seasonality = "auto", weekly.seasonality = "auto", output="model")

  expect_equal(last(model_df$data[[1]]$`time stamp !"#$%&'()*+, -./:;<=>?@[]^_'{|}~`), as.Date("2010-01-23")) 
  # Make sure glance_with_ts_metric works with group_by including special characters, especially pipe '|' characters.
  ret <- model_df %>% glance_with_ts_metric()
  expect_true(all(c("RMSE","MAE","MAPE (Ratio)") %in% names(ret)))
  expect_true(!is.na(model_df$data[[1]]$forecasted_value[[length(model_df$data[[1]]$forecasted_value)]]))
})