tsibble - SKILL.md Agent Skill

name: tsibble description: R tsibble package for tidy time series. Use for temporal data structures with tidyverse integration.

tsibble Package

Tidy temporal data frames.

Create tsibble

library(tsibble)

# From data frame
ts_data <- df %>%
  as_tsibble(index = date)

# With key (multiple series)
ts_data <- df %>%
  as_tsibble(index = date, key = c(region, product))

# Regular interval
ts_data <- df %>%
  as_tsibble(index = date, regular = TRUE)

Index Types

# Yearly
yearquarter("2020 Q1")
yearmonth("2020 Jan")
yearweek("2020 W01")

# Convert
df %>%
  mutate(month = yearmonth(date)) %>%
  as_tsibble(index = month)

Temporal Operations

# Fill gaps
ts_data %>% fill_gaps()
ts_data %>% fill_gaps(value = 0)

# Filter by time
ts_data %>% filter_index("2020" ~ "2022")
ts_data %>% filter_index(~ "2022-06")

# Lag/Lead
ts_data %>% mutate(
  lag1 = lag(value),
  lead1 = lead(value),
  diff1 = difference(value)
)

Aggregation

# Temporal aggregation
ts_data %>%
  index_by(year = year(date)) %>%
  summarise(total = sum(value))

# Group aggregation
ts_data %>%
  group_by_key() %>%
  index_by(month = yearmonth(date)) %>%
  summarise(avg = mean(value))

Rolling Windows

library(slider)

ts_data %>%
  mutate(
    ma7 = slide_dbl(value, mean, .before = 6),
    ma30 = slide_dbl(value, mean, .before = 29)
  )

Scan for Duplicates

# Check for duplicates
duplicates(ts_data)

# Check regularity
is_regular(ts_data)
interval(ts_data)