NEWS

gseries 3.0.3 (2026-09-01)

(G-Series 3.0 in R)

Compliance of tsbalancing() with OSQP V1.0:
- Setting polish renamed to polishing.
- Change of OSQP object: from a R6 object of class "osqp_model" to a S7 object of class "OSQP_Model".
Correction of a minor bug in tsbalancing(): column require_polished was missing in output data frame osqp_setttings_df
Minor corrections to the documentation (e.g., fixing typos, reformatting the code in the examples).
Awesome Official Statistics badge (README.md, index.md and fr/index.md).

(G-Series 3.0 in R)

Minor changes to comply with the CRAN requirements following the second submission process:
- New package title (DESCRIPTION).
- Return values added for function aliases (aliases.R).
- Operator <<- replaced with environment specific assignments (for more predictive behaviour).

(G-Series 3.0 in R)

Bug fix in tsbalancing() to avoid altering the R session's error option (getOption("error")) after execution.
Validation of class "data.frame" input objects in (relevant) utility functions for coherence with the core functions.
Updated description of the function arguments regarding the expected class of input objects.
Minor changes to comply with the CRAN requirements following the initial submission process.

(G-Series 3.0 in R)

Initial release of G-Series in R (package gseries).
New functionality (stock_benchmarking()) designed for benchmarking stocks using a cubic spline interpolation approach.
Improvements to the already existing functionalities:
- Benchmarking (benchmarking()):
  - new options allowing proportional benchmarking (lambda != 0) in presence of negative values in the input data;
  - a flat indicator series of all zeros is now allowed with additive benchmarking (lambda = 0).
- Raking (tsrasking()):
  - optional alternative handling of negative values in the input data (same handling as tsbalancing());
  - helper function (tsraking_driver()) to simplify the reconciliation of several periods in a single function call.
- Balancing (tsbalancing()):
  - customizable solving sequence (multiple attempts at solving the problem as opposed to a single attempt);
  - improved solution validation process with more information (output data frames) available to help troubleshooting invalid solutions;
  - optional input data validation only process (without attempting to solve the problem, i.e., resolve the discrepancies, if any).
New utility functions to help using the core functions, e.g.;
- for producing benchmarking plots;
- for converting reconciliation problem metadata (tsrasking() to tsbalancing());
- for manipulating time series data (prepare or convert the input and output data objects).

(G-Series 2.0 in SAS^®, contact g-series@statcan.gc.ca)

(G-Series 1.04 in SAS^®, contact g-series@statcan.gc.ca)

Name change: Forillon becomes G-Series.
Introduction of alterability coefficients for PROC BENCHMARKING (when rho < 1).
New WITH statement for PROC BENCHMARKING.
New options WARNNEGRESULT|NOWARNNEGRESULT and TOLNEGRESULT= for PROC BENCHMARKING and PROC TSRAKING.

(Forillon 1.03 in SAS^®, contact g-series@statcan.gc.ca)

(Forillon 1.02 in SAS^®, contact g-series@statcan.gc.ca)

(Forillon 1.01 in SAS^®, contact g-series@statcan.gc.ca)