NEWS.md

In timekit: A Collection of Tools for Working with Time Series in R

timekit 0.3.1

• Improvements:
  • tk_index and tk_has_index: Added decomposed.ts
  • tk_get_timeseries_signature and tk_augment_timeseries_signature:
    • Added "year.iso" (Issue #4)
  • Cleaned up Imports:
    • Moved tidyquant, broom, forcats, scales to Suggests
    • Removed quantmod, TTR, timeSeries and tseries from Imports

timekit 0.3.0

• Improvements:
  • tk_make_future_timeseries:
    • Improved future date picking algorithm to look for seasonal trends by way of inspect_months. The data must span at least two calendar years to gain this benefit.
    • Improved handling of index in situations with units = days and scale greater than day (e.g. weekly, monthly, quarterly, +). Now values returned will be a regular series of the appropriate scale versus an irregular series indexed by regularly spaced seconds. For example, monthly dates in units = days will be predicted as the first day of each month rather than by the median frequency in seconds.
    • Includes a new argument, insert_values to add time-based values into a time series. Before you could remove via skip_values, and now you can add via insert_values.
    • Changed n_future to be inclusive of skip_values and weekends, which means that the end date of the future period is the same regardless of inspect_weekdays, inspect_months, skip_values, and now insert_values. Now the end dates are fixed which helps when manipulating future days.
• Documentation:
  • (New Vignette) TK02 - Making a Future Time Series Index using timekit
  • (New Vignette) TK03 - Forecasting Using a Time Series Signature with timekit

timekit 0.2.0

• New Functions:
  • tk_make_future_timeseries creates a future time-based index from an existing time-based index.
  • tk_get_timeseries_signature returns a tibble with the time series index decomposed into year, quarter, month, etc.
  • tk_get_timeseries_summary returns summary metrics for the time series index including number of observations, start, end, units, scale, diff summary (summary for frequency in seconds), etc.
  • tk_augment_timeseries_signature adds the time series signature to a tbl (with time base), xts or zoo object.
  • tk_get_timeseries_variables returns the variable column names for date, datetime, yearmon, or yearqtr variables in a data frame.
• Fixes:
  • tk_index: Fixed timezone issue with yearmon and yearqtr classes.
  • tk_tbl: Changed argument from index_rename to rename_index for consistency.

timekit 0.1.0

• Initial release of timekit, a time series toolkit that simplifies coercion between time series classes and time-based tibbles ("tidy" data frames).

timekit documentation built on July 4, 2017, 9:45 a.m.