create_gran {gravitas}

Build dynamic temporal granularities

Description

Create time granularities that accommodate for periodicities in data, both single and multiple order up. Periodic ones might include time granularities like minute of the day, hour of the week and aperiodic calendar categorizations may include day of the month or week of the quarter. For non-temporal data, supports only periodic deconstructions.

Usage

create_gran(
  .data,
  gran1 = NULL,
  hierarchy_tbl = NULL,
  label = TRUE,
  abbr = TRUE,
  ...
)

Arguments

.data

A tsibble object.

gran1

the granularity to be created. For temporal data, any combination of "second", "minute", "qhour", "hhour", "hour", "day", "week", "fortnight ,"month", "quarter", "semester" or "year" can be chosen in the form of finer to coarser unit. For example, for the granularity hour of the week, value is "hour_week".

hierarchy_tbl

A hierarchy table specifying the hierarchy of units and their relationships.

label

Logical. TRUE will display the month as an ordered factor of character string such as "January", "February". FALSE will display the month as an ordered factor such as 1 to 12, where 1 stands for January and 12 for December.

abbr

logical. FALSE will display abbreviated labels.

...

Other arguments passed on to individual methods.

Value

A tsibble with an additional column of granularity.

Examples

library(dplyr)
library(ggplot2)
library(lvplot)
# Search for granularities
smart_meter10 %>%
  search_gran(highest_unit = "week")

# Screen harmonies from the search list
## Not run: 
smart_meter10 %>%
  harmony(
    ugran = "day",
    filter_in = "wknd_wday"
  )

## End(Not run)
# visualize probability distribution of
# the harmony pair (wknd_wday, hour_day)
smart_meter10 %>%
dplyr::filter(customer_id == "10017936") %>%
  prob_plot(
    gran1 = "wknd_wday",
    gran2 = "hour_day",
    response = "general_supply_kwh",
    plot_type = "quantile",
    quantile_prob = c(0.1, 0.25, 0.5, 0.75, 0.9)
  ) +
  scale_y_sqrt()

#' # Compute granularities for non-temporal data

library(tsibble)
cricket_tsibble <- cricket %>%
mutate(data_index = row_number()) %>%
as_tsibble(index = data_index)

hierarchy_model <- tibble::tibble(
  units = c("index", "over", "inning", "match"),
  convert_fct = c(1, 20, 2, 1)
)
cricket_tsibble %>%
  create_gran(
    "over_inning",
    hierarchy_model
  )

  cricket_tsibble %>%
  filter(batting_team %in% c("Mumbai Indians",
                             "Chennai Super Kings"))%>%
  prob_plot("inning", "over",
  hierarchy_model,
  response = "runs_per_over",
  plot_type = "lv")

# Validate if given column in the data set
# equals computed granularity
validate_gran(cricket_tsibble,
  gran = "over_inning",
  hierarchy_tbl = hierarchy_model,
  validate_col = "over"
)