actxpspy v1.1.0

• expose_split() bug fixes:

  • ExposedDF.expose_split() was updated to respect the values of start_date
    and end_date originally passed to the ExposedDF() class.
  • Future policy anniversary dates falling on February 29th leap days are now
    consistent with ExposedDF()
  • New tests were added to verify that the sum of policy year exposures
    (exposure_pol) after calling .expose_split() match exposures produced by
    ExposedDF.expose_py().

• ExposedDF() bug fix - quarterly and monthly calendar exposures periods now
  strictly calculate exposures based on month-end dates. In the prior version,
  months ending on the 28-30 would use that same day of month for subsequent
  exposure periods.

• ExposedDF() and ExposedDF.add_transactions() now allow date columns to be
  passed as strings in YYYY-MM-DD format. Any strings are converted to date
  behind-the-scenes, and any missing values will results in an error message.

actxpspy v1.0.0

• To improve speed and efficiency, the data frame backend was changed from Pandas to Polars.
  • The data property of ExposedDF, ExpStats, and TrxStats is now a Polars data frame.
  • ExposedDF, ExpStats.from_DataFrame, and TrxStats.from_DataFrame will accept both Polars and Pandas data frames. However, Pandas data frames are immediately converted to Polars.
• Removed all add_* date functions which are no longer needed under Polars
• ExposedDF's groupby method was renamed to group_by.

actxpspy v0.0.1

• Several updates to align with the R version of actxps

  • Added optional confidence interval outputs to ExpStats and TrxStats objects
  • Added an expose_split() method to ExposedDF classes, which divides calendar period exposures into pre- and post-policy anniversary segments. This creates a SplitExposedDF object, which is a subclass of ExposedDF.
  • Added special plotting functions ExpStats.plot_termination_rates(), ExpStats.plot_actual_to_expected(), and TrxStats.plot_utilization_rates()
  • Added from_DataFrame class methods to ExpStats and TrxStats objects that enables the creation of experience summary objects pre-aggregated data sets.
  • Added the great_tables package as the backend for table() methods
  • Shiny UI updates
  • Added a 'points' geom, log 10 y-axis, and confidence interval options to plot methods
  • Added sample data for pre-aggregated exposures (load_agg_sim_dat())
  • Bug fix - the target status wasn't being fully exposed when using calendar period exposures
  • The assumed default status on ExposedDF objects is now the most common status instead of the first observed status.
  • Added helper background functions for column selection
  • Added several articles to the package's website

• Current differences from the R version are:

  • No support for plotting a second variable on the y-axis for .plot() methods
  • No predictive modeling support function (add_predictions(), step_expose())
  • Table output cannot be saved to a png directly in the shiny app
  • functions like pol_interval() don't accept arbitrary durations. Only 'year', 'quarter', 'month', or 'week' are allowed.